From d7acba20bddbba5df147c806df2deb545c251633 Mon Sep 17 00:00:00 2001 From: yanirs Date: Mon, 10 Jun 2024 04:24:19 +0000 Subject: [PATCH] deploy: a299cdf49058f97fcd1a920b0488597aaccd5311 --- .../index.html | 20 ++++++++++++++++++ ...kafka-inspired-startup-data-pipelines.webp | Bin 0 -> 121924 bytes ...f88_121924_1080x0_resize_q75_h2_box_2.webp | Bin 0 -> 94984 bytes ...5f88_121924_360x0_resize_q75_h2_box_2.webp | Bin 0 -> 15454 bytes ...5f88_121924_480x0_resize_q75_h2_box_2.webp | Bin 0 -> 21714 bytes ...5f88_121924_720x0_resize_q75_h2_box_2.webp | Bin 0 -> 48854 bytes index.xml | 2 +- posts/index.html | 2 +- posts/index.xml | 2 +- sitemap.xml | 2 +- tags/analytics/index.html | 2 +- tags/analytics/index.xml | 2 +- tags/business/index.html | 2 +- tags/business/index.xml | 2 +- tags/data-science/index.html | 2 +- tags/data-science/index.xml | 2 +- tags/data-strategy/index.html | 2 +- tags/data-strategy/index.xml | 2 +- tags/index.html | 2 +- tags/index.xml | 2 +- tags/startups/index.html | 2 +- tags/startups/index.xml | 2 +- 22 files changed, 36 insertions(+), 16 deletions(-) create mode 100644 2024/06/10/startup-data-health-starts-with-healthy-event-tracking/index.html create mode 100644 2024/06/10/startup-data-health-starts-with-healthy-event-tracking/kafka-inspired-startup-data-pipelines.webp create mode 100644 2024/06/10/startup-data-health-starts-with-healthy-event-tracking/kafka-inspired-startup-data-pipelines_hu153a73840b0658eeafb17cde694e5f88_121924_1080x0_resize_q75_h2_box_2.webp create mode 100644 2024/06/10/startup-data-health-starts-with-healthy-event-tracking/kafka-inspired-startup-data-pipelines_hu153a73840b0658eeafb17cde694e5f88_121924_360x0_resize_q75_h2_box_2.webp create mode 100644 2024/06/10/startup-data-health-starts-with-healthy-event-tracking/kafka-inspired-startup-data-pipelines_hu153a73840b0658eeafb17cde694e5f88_121924_480x0_resize_q75_h2_box_2.webp create mode 100644 2024/06/10/startup-data-health-starts-with-healthy-event-tracking/kafka-inspired-startup-data-pipelines_hu153a73840b0658eeafb17cde694e5f88_121924_720x0_resize_q75_h2_box_2.webp diff --git a/2024/06/10/startup-data-health-starts-with-healthy-event-tracking/index.html b/2024/06/10/startup-data-health-starts-with-healthy-event-tracking/index.html new file mode 100644 index 000000000..19a890798 --- /dev/null +++ b/2024/06/10/startup-data-health-starts-with-healthy-event-tracking/index.html @@ -0,0 +1,20 @@ +Startup data health starts with healthy event tracking | Yanir Seroussi | Data & AI for Startup Impact +

Startup data health starts with healthy event tracking

June 10, 2024
Kafka-inspired startup data pipelines

The first question in the Data section of my Data-to-AI Health Check for Startups is:

Do you track Kukuyeva’s five business aspects as wide events?

Whole books can be written about this question, so I figured it’s worth its own post. Let’s dive in.

Why track and when?

In my past decade of work with startups and scaleups, one thing has remained a constant: There are always gaps in the data.

Data gaps may take the form of missing data or low-quality data (from partial to deceptively wrong). The feasibility of any AI/ML automation or data-informed decisions depends on the nature of the data gaps.

Another important feature of data tracking is: You can’t go back in time to collect or fix most proprietary data. In other words, historical gaps in the data are there to stay.

Therefore, the answers to the questions from the heading are:

  1. Why track? To make data-informed decisions, and support automation and optimisation of decision-making via AI/ML. The latter comes after the former, as you need to turn your data into trustworthy business metrics before you use algorithms to optimise for those metrics.
  2. When do you start tracking? As soon as you have interesting data to track, along with the commitment to ensuring that you’re not just tracking garbage. Typically, you should start once you move from a throwaway prototype to a product that has user traction.

What are wide events?

An event is essentially a timestamped key-value document. Wide events encourage liberal use of attributes to support downstream exploration.

Here’s an example from Ivan Burmistrov’s post on the topic, of a wide event for ad impressions at Meta:

{
+    "Timestamp": "1707951423",
+    "AdId": "542508c92f6f47c2916691d6e8551279”,
+    "UserCountry": "US",
+    "Placement": "mobile_feed",
+    "CampaignType": "direct_ads",
+    "UserOS": "Android",
+    "OSVersion": "14",
+    "AppVersion": "798de3c28b074df9a24a479ce98302b6",
+    ...
+}
+

Burmistrov’s post is worth reading, as it delves deeper into specific examples of how such events are useful for observability and exploration of unknown unknowns: things you weren’t aware would be interesting when the code emitting the event was written.

Having been on both the producing and consuming side of such event streams, I have a few thoughts to add:

  • A common objection to Burmistrov’s post is the cost of tracking. This depends on implementation specifics – there are ways to balance cost and usefulness of events. Further, if a startup is growing and more events are emitted, revenue and funding should also grow – which supports higher-volume tracking. Further, tracking costs should be compared to the opportunity cost of data gaps and to the cost of engineering time for cost-optimising the tracking system.
  • Many software engineers don’t think in terms of events, as data models for production systems typically reflect the current state of the product rather than its entire history. This is unlikely to radically change, as this sort of data model makes sense in production. To bring engineers along for the event ride, it’s worth getting them to read the classic article about The Log by Jay Kreps.
  • As a step towards data quality assurance by event producers, automated tests should ensure that events are emitted as expected. This is different from traditional logging, which is seen as a side effect of the system that doesn’t require testing.

What are Kukuyeva’s five business aspects?

I came across Irina Kukuyeva last year, when I was looking for other people who are offering Chief Data Officer engagements. Her website is a treasure trove of advice for founders, investors, data consultants, and others.

On the topic of data that should be tracked, she has this to say:

After 10 years of collaborating with companies of all shapes, sizes and industries, I’ve found that all aspects of your business fall into just 5 attributes that you should (legally and ethically) track across your platform/hardware/service and your customers, timestamped at the individual event level:

  1. “Demographics” of the customer’s/IoT device (e.g., Android/iPhone, tablet, desktop, hardware type),
  2. Your app’s/platform’s “demographics” (e.g., release 0.9.3, pricing plan(s)),
  3. State(s) of the app’s/platform’s assets (e.g., inventory, sensor(s)),
  4. Interactions of all of the above, and
  5. Touch-points (e.g. acquisition channels, marketing emails, customer service interactions, sensor maintenance)

One thing I found confusing when I first read the article was the use of the word attributes, which commonly refers to specific event attributes. However, now I understand that not all aspects are tracked in each event – it’s a broad overview of what should ideally be tracked across the business.

Importantly, when thinking of tracking at the logical level, the tools don’t matter. This is exemplified by another Kukuyeva article that advises founders to get started on their data strategy with tools that can be as simple as pen and paper.

For examples of how events fold up to analytics and other use cases, see Eventify Everything by Timo Dechau and Activity Schema. However, keep in mind this observation by Misha Panko of Motif Analytics:

It took the world decades to develop widely accepted standards for working with relational data and SQL. I believe we are at the early stages of doing the same with event data and sequence analytics. It is starting to simultaneously emerge in many different fields:

  • eng observability (traces at Datadog, Sumologic, etc)
  • operational research (process mining at Celonis)
  • product analytics (funnels at Amplitude, Mixpanel)

As with every new field, there are a lot of different and overlapping terms being suggested and explored at the same time.

In short, even though systems for event stream processing like Apache Kafka are now over a decade old, the industry is still figuring out how to best model and use all the event data. No one has all the answers, and your mileage will vary.

What’s a healthy level of tracking?

Assuming compliance with relevant data privacy regulations, there’s still a question of what tracking level is healthy. Despite the yes/no phrasing of my original question (“do you track Kukuyeva’s five business aspects as wide events?”), responses are likely to fall on a continuum between 0 and 1:

  1. Obviously unhealthy: No, we don’t track any events.
  2. Unrealistically healthy: Yes, we track everything and have high confidence that we know everything we need to know.

Health is relative to the stage of the startup and its current goals. If data gaps are blocking growth or likely to start hurting growth in the 6-12 month horizon, then tracking is insufficiently healthy and should be addressed.

Data-to-AI health beyond event tracking

This post is part of a series on my Data-to-AI Health Check for Startups. Previous posts:

You can download a guide containing all the questions as a PDF. Next, I’ll go into the other questions in the Data section. Feedback is always welcome!

Subscribe +
    + +
    Alternatively, subscribe to RSS feed.

    Public comments are closed, but I love hearing from readers. Feel free to +contact me with your thoughts.

    Text and figures licensed under CC BY-NC-ND 4.0 by Yanir Seroussi, except where noted otherwise  | +Powered by +Hugo & + PaperMod
    \ No newline at end of file diff --git a/2024/06/10/startup-data-health-starts-with-healthy-event-tracking/kafka-inspired-startup-data-pipelines.webp b/2024/06/10/startup-data-health-starts-with-healthy-event-tracking/kafka-inspired-startup-data-pipelines.webp new file mode 100644 index 0000000000000000000000000000000000000000..95a2488386d4bf3a50da3e85847cdb9c0a2e7c51 GIT binary patch literal 121924 zcmaI6V{|1^_wKu6v(vF{Cmp9_+qP|I$F}VbI<}ooI<{@w+_&HN+;RWs!#Q`2v1dc@4KZOwbww^scmM!E{BPDm0^Gm>QX(RX1>pZ~0pJC;CU!26 z%m9F$y{ogbgfOwDmNqfW2>==Z4}bum0I-crTpWcI6=naE{eK=$O8~I{MCKX(v)BJQ z=l`lAnVPwn003ab|1y4>I6AxjW5a*U>EY`5AI|v47$%lRrvJF=AJaSkJK#TF`Hyb= zzxd)mZ1G?IZ&4V|s>&k&YQy-a6I=XmZ2Z5miKVmMzZ~{|Ipn5x_W$+=*Zf~>`XBcF z58K(g{X5%#$A2nBG_zMz`L~k)oA>|;fD}Lup!lz?#sD{f6~GqY3SjuR+W%8r0LuU3 z!v7aP?tlG?|6IoZTvh;+e-2TA1HcYo^j|;VKkwjQ&cE?Ly>&5XWBsoLhA0dGKy86Q z_Y?pCOfmrQnE(R47lJ^aMF0TA1_01!|37|*0sw&P@t>aXKN@*H0Duw#0JIGKk7k?> z0JMe!0NAUJM$Sh6eGbIGE4aBi0B~0Z03c}r0N-W+0C=7M)wh4K|HJ{s%K(7tzh22s z0sxs=005=MzrA(;FLlHIGyH#j`~S`PfBX0E%O!SS1p_27LFaLP( zlF+K~GuNTRnHif^^q?RL68JZ#O;0ZaTRze54ZgMt*IO5Q7s~vZ7<#o%6G7<(g1><$ zk4_)cvl2ju!@Z$=py3C{80{KkXUA9Pn_9oYh4%An{?qic?*P~Jy}>-G%h z0eA|$1NHg>fQ(;dZ=j3!j~@kCC)?}2mI44^3Q+5x@p>KR*`T$@1!x44de^xkda?mI zyrI}#TW)V6*>|X^J`qBVXZP4)LupAe67^w_3w%bAe8f$_MZrWrN9IVFh%2 z%5+_=4QjPG>bx|xJ)3Vc_l3I7%0r{jG-VW)>#+pBBPU4q)B(V`ud z-ZAM7y(m^56}%%=77=?wo+G}&zJeCz3D=&>2~Md+8DW&OC46Xz~E{&Z7mTZM)I2M_0@^i)RUnpUjb@lws!7b~*>uR4wfCGi7-cqdQaxtWT6DZqU zQ^j2z8yS8#!qIr@+uzdlWsr<5ZpAI286BiCSel<7S=j@u0#Q_LGR$;HwF;x!8a2!H z+Xbo0JQ54cIb1^Jlr9GD#u30;tS?S3W_S$1i}nZ1A1`%7c}z|~-ewv1&suj*p?@Dx zHu2W(B)$u~yz>8n-TDyArj_!~( zM(Xf6q*bgzqa~U%t0xC`IK)YMjx+?^7$T6(*rUo;uW2{-FGFuuQ+xZJh&wA3C&u;% zC8Qja76Q|P$kSRA#M2nyWo;m|{{ibGpnkWGS3%ob%7AE#cPvEABlj(NAjkoDa8G`U z&&Mh(Kf)bOWpM#L){)E*tZEJkXZN*sif2p79oJiC2v0UGe;qZ8RQ^G@p4Y@aR{^Q# zFl@umm+x1~aUl{DsB|g`s1qc0mvF%p)N%B!pL6&_r&GD;X$$PkOeS*gVZ9NO)a#1k zdi&FA$$Q#UfBC&3qB}&GDyK^6?6{7_-QjSwX?Rz+K5Bfu+32kS`yrU`L^zEjzv^lS zD?N2xCQU_R^4I%~a7>s|kl2YxzYL&_Ge6w;yh+S$3JdwiQpByDu`}v41U`qpi_l+w zX;pbJZv$oXZz0#movoPtLNZ{fzpRo&>RYuG_vOc62(+@cp0A+0ChtgINt zWO4UQXZ+nH^A=GeiWaZ_!Q`NZ9hQMl!nW2$_r-PXAQy}+2H=KuIY%n)(d`hLDyq4= z^Pts387U#xeiTzXF<*ieiThD>%}!ySw=6$j#r?RH*oN=fn5PbhZJf4{IUFG2Fh16@m(c zRj_|}{Kz6BHYak%e9>Y7{iV4WhM``$$ahj-Je|UTHP>w2PkoV{8jqwk+jWMpF&#%M zw+IB>Sc?*Y8#cOUh(5Az^iz^hGQI7}m=$#Sk!Cc2>$J<7(^)W!QN#2lL+8L#3w=%j zKj2&}RmiUC(2tDI4Og`f=k==3PRX0OwQmSI6_w`9HazA5`wC4;$8DINXY}H83Q}mh z&}-p3yWe`Bew&dBjAzx==<*Dh`HU3=Xj{$2`IxJ{BRe#o6yursZN_8R&-8G4u52M; zE7B4lkl~kwVTZG35&hKvty1hKm&qU~SB9ye6}$2bIl5a{JGPV$`;O8Cq3=6N!*9B- zDHpU+L-Y9^!*TQJVc%{mxshsS!1}S>r6nNavP6~XDr|eRY`b+nr+m^H zjnbrD;BXMe_hm%IY$vXK=(yVK{b@%1H9=F>D*=VodQ#a=8*2878#?-ihKC49+ombH zSj%wR1MYNfoF+etnjY$;5jHo0>vq#MtU&Kpf1vO&FWfK&6c^}N5fMbyIlbUK;mr)$ zuP=FLi`S38e}6C7(_3HYTGA9yFm9B#0Y}(K5B1FY!8b?nF%XBLo%qUyL^$aN}43-y0^P2}hR%r-r|X(QoTqM!6ak z;+8WiCxHwdFN*EOO~=Tu8@TtX4}T~dpbYhce#P7lAlC-9DB7)wli&1Yex?UCX8PEm zVtkr3A7?D5yZWZomm6##wGx+`12Mah3WB1*2s&Q@TMPQ&0rhiFW(3}HANPdr&3ay%^17?9odRnA}t#Qv_J6`AAbEB_ROP$8S#0P{*5UbI(6H+#RA_-5& z?l_q*0q1!jKv_=aQ(aa#cB!pDYfVc=o5t~Utuq~KLv>ewQ&O+Nxz+38L+(pb^yp3` z3;X5A2)V_%ax%kWlVGah?oJ(n<@yavm7ZC*IST^XW%a8!qdFwk&L^_GuUbCQV~fk? z38}g@nkM7gT(D%SkhRnv!d|kd-ieg~u{0%Ue)O#o3qTDfp{mA=mx_L|=d7gi1zDiJ zwi%zvP3Q8%F}g+RrLToS{MiMX6Gb0^X>7JArI3FFhJ`HU6bS~@um%-ur~VsjO^18u zp2)jc;Fjenu=l%eIxR>s51vyw}k zkAnL$GbG-Y#0r^HYV|iopxdjY{f`t@c$ai~-(`0m_?br8WS5{#3A1xb{IyA^2ouzi z5&VI4a0{0l=q{2j7!A$Rk`ldnCkmbZ(mg&vJl}2(p9vdxAnio8O75z~Fn@f7r17+D zIQqJjBWo2zM>Am>tkJi@V%o7l9~^ygj%rmk{cwGqN}NAah?ci43+JGV$}T!qYr5nl zuUTb_TbOCd$Dpdzs&7~cpN0f=p$&An-Yd>I5hSRBz0oIP%{3N^nJ5AxxAd6l(#Hc>+o;xj96_4$?jg`(Ub?sgC!JClNx~NKfua zyY=kfEN7i#-ta1i&7$8{9n_il3j=rBa`w1(>>}_ibtCKicBy{m2=d;f931yi4jzMc=Y_jrqa#rg@58Q+3T`A?9mIuU&b z$MGKTE=qIam{PPH%3;RG+XA=>9G0!C4SJR*838t&aB_o zpG!;axXsKB1$|mWDY)uNBZd@R;RwHle%;aZ}h|uK)Uyx-ds7` zzAf!e#wa}Z_yw7|sgS0(B8l*41&sDW$(7H9Wf+R~58urikcXHk9-+*)n@viA%LUOF+W(a>{w2$yLWg)f0-Hbr}361#10Ou@uM)t_TIyByL z!GBNM=|S4`^PUJhAMteKWF?!0k@puY-pp>{4eoVzIZa6efGak}J1<_WZjUVuDu6+L z;LZxn+;{1NPuc)k6wzwQ*x;IZ*cu`sh44>isIV-$Mb6?`R4%R!ewa6wB;N+>HiD9a zf)2jBvbkYdx0nc?$a0t9LkV$c2w9>M}w;iCK_lpe6RLgaF*-8S!L;HNS}f<~|%nUjj?$u9ZmfmKme@lGGdwBqz<=7FlK z6cK@Rz#F|$tjBO5M4yQY^P-jd$+A)A#_9-4<%U|X=l3e6vwr86U||6rx9mg?W#Zl5 zd79~vNK?f?Vw1Ox#o)Q@ekpDwDE_}+Ie&8YVO9~4-<|mWw$?O~!tD>RMpS+c_D7p0 zCVyJ_dkIX%Ys(@oeRt4eiW{LA1VWXGP3HCmBEBFt_myz5jmd<=-#tx}&_R&9i1;YIg+O73%@TCI`^UPNLJUo9BMKBFTr`Q=Bo zL735>$%zkyI{G2o^Xslln(tAl28xXUmFlRS1M#rtByw?{(F!g*MNW2(?!;Gaw=q0) z=%|s9P_I#V2-%;4lxRd9zNxXsIG22E3*fPA)l3@n_#3$;cj>$7!%hj{&{%F0ZraIm zH-_ciqoL>xl$N8+ z(j#4N0L0-swnpb7Op2BKpz7#ut?6L7O6VhlNm~NIqj{qHI}^mI4uYR4F_F3>q}y`7 zq)Z@eME-KyBEuOJQiBdcG4oykOzeZm<=f2CQT!H#|Z;zGsCL zBs*xG_fEl`kb7GA;q!hkBhdocm?P0{+{3?z6a?cdtz&+N-DeoxaaFb?JaX*VtepVZIru^VUTZYQv~q zsb`UY7&alQsYzmXkcC*upWD0PGz;)XBCr!6BF|Q;t_ZTBq+xW12;hE*BZHruYBA#IV-@KNMT-GKf-v{1Os7PY}M&^`ij&B%n!4{$J zPLw}ZBaJs*<~ibhA&TLKrw%%(?`C;dyf20aoD!JUjz6f z=U%RYPj0GRlU~A}ggYt=dTPjiRm7T$bK9(cs^R}Gm}!(>mjiS_6oya|!MqqHIHCan zTVO?YV>L?Kx_CvXFCJ8R2rp>Y(oW)#Y>+qR7Wl}w(@Yu*_PO+}z$Hu@u#pg-KQ}sq zMjK8mcAG7xFz(NPmoFTS*&X<3@)IP7HyJW4pw}S}{>a-ut;>Af+`9)1b+o60t{-J* zz|+E*+(NuJPOosjpe1ar%F*9H7{y#?T-cyEn9R0C|8NYOeH*OW;wn4tFY);P_qcFD z`5DIC5LGhqFkbahQ~uZoBTS>NnMGV>jCl7)!;)iR$#?CNbnvuaU95*&6Ov0mDzMe% zd`A$X&S^r!@P6<8VwAq=xV}d8furu#LqYCQP=$YgpO`Lb(gXubrrReY8uKF zM(<#6KgGA5GMn-+=_fdXxN@#k;ShZjnRu66(xa%l@*`@!tT+qn^~tv!RcW+Km*8zs zKi^quj?CcF>jWp$Em=R61=NoZ#P_W2diRz3q>LC6orQp(P4-9HEDPVdPP*mB!+bOF z*z=yP#ON79ak&n%L-ZoRO@LE&x5;1MYQ2CyoTYF2qP%pQmkV<2A_mThGxp1m3K@szBtPL#^ha#i z458aE;)z$M)2gnT@Aqwiu~!D^881|}rF0q0+h!tzmq0vIQpd_14vPl%?6y45Bw7st zcO1FQ%wvVgj=RC1kn`6&9*#<67L8!31u%}D_yL=_Ff)W8KRrf63GRMa2ZD{J+DFYYIwiutkAF2G1)*o%i_sN={_ z(2>o(^HQJa3SSC;z1-R0%L7f7GW`O|x$=LcB1HrxfNjla$Vy47b-LNBpy7_`Ypx|F zS3XS15%dsDYne*vak25>NYZ;h@x{oBLYxq2tGu-vGI!M0n78_Ud)hA+N%S2ssm~!s z42soj9}Ovt&7BxE&tGrfsjPl208i6D=H@3aRY33(HDy-j<@yS*2t!;x6nrshxpK`e z(Pq*foRcZ&0~fY5kr+2_1y6v?!^KxuMl&ydxnx{PZ5%RelH>&AAL~8*a@$nt{=50q zI;kC}?`sW~rDLg{kZ$2LBqpw^huM#V^dy`V#VJ=92rFTV&BzZ!e{30(nGjSRyO%Y? z<=^uPk+hjH!^r^EA9(lI?V1{%TL_A-L(+gE>LZiska2cbtEE>QzH{=!&Z_!w3aO5N;m=ggT^}Kha0Z*6x=`V&pdW#zq zX1lQYZG$qP*EMvmwMqC_>p6-~_FhIIB*nmyAGBqdpvy(fdVYJOkUKQX&@8S!{0}Fn z+23Me?f&QU87>rijI378$^_u8v@x%9K;SJ)8M*atQdU0C#~^HHiGKIz1W}zl z;)f&@s%0BWDQ$xZLuniFKA^jNo`xmfn$`DgZG&lbeKh@9lydKGWUzOJ1 zDc?`BXB@C0<1KknE~OcbC1osV?(BrO;CU z*kGWaK75ovn_K?dY%U)0Y#CFtin(Hvi}lN%bw=4N_gqp{0mr$1xslw0>8#0dw+ssR z^!+yBZd+>KNr)?d*vSMua->d{N-18h5@{aBw7iEX=lKPbT+l^G2hKp!85uj)rzk_g0L>T^&0dY$=xHgz#s= z3{%of{Ax|`Fvn*=I#!a#%5D;hX)7})K0?)ObWOH?i@b;OXr?&aW%`ZBn7a!_>n|^v zSOVG-!%h4xK4LQCB#N66Z=R>*vCE8esB&L!2b?uM&i@5=h1Q|p0!@}otLE{M1nC0Ld2(I|LVk(jXo^w> z@(CmiQl3T(vZpe-BszX_%xB`2_jTX)gRyqZa#2@Ltu0_C&=o^H>tgBECDct3LyX4Omev=m10?nNZ&j%>ARg_E~&*2usZhc0A? z%}7?}LjBRI3jb1hHJ^Z2QVucQrP#*{6@Cqi8=kD?B!!;5WV~Z?zTT{MRvXhJEH(9e z(4_U3xN|qC0OB!TV(lPT?l8ono(RAiyDKQ6PEVEboPJ>+Vby)TTw z4EbBn41rwq${aRwgP{^(Xw6B|fesO4FzKEC(E%5+xKV!FG}B4^s2iN=spP{yKt+Bj z@_8?-5Rt90-VMbDhF=I6eYEn!^ma0N9Pc=~1r8@TE5LKH^C9c67MVS4$^kOSj<3=p4kobx(NRKHqRR_Ir!uB|!=_pF%* zaMNHzX?nq*j-40(9XLwlweZ|&lN4fYvOZJiTaqi!g`$&hCuENC?e1z|6(V6@oB;nJGtwp zlK@%=w5y#lwG8p6FErF7S7PgV@7V|E#F!`jG%_4d({oV*pP7gcPjwqxh;}0GAoX3X zd_MtO{53|`iO&%A=%YN>9_sr!19^LN?w?%I-ONo0BiSF!xBF9>_0K|yrG<}yo+4ef z)+x8^4#jOJ=&^H{(`4uM>5sXOPGETmBb%Z#_YjbttgGX>K O_nDKX>swjNn`ays zNIHBUi6$<~zz!GK($W<^q^URm3-om?D>&NjkP7sEnYN>)qYTGClV#^#B!r|Z)hA6t zXPDecj9g{9QO)Lz5LJ5o&3Id)$#@BYK9a?Y`uz~L{5^3=XqDBi8FztnTu4`G*kaJg z2*Zu->DKV!&tH>yryLA{PET2eUzucq??3HC!b|HbHq=Oi#jZ7Raz4-YCObG??#4h+ znSQh6%w{ek>N29dSy|mTlhFO)u6YjDmV@W4%9FJc?XDlnw4;&}#{_2AxobAz_Hgw#ap!J2%1F z1Sae$EL(YwdS6(j2E8pCoULA4XjQzI*AQk}o|h$$-0G#NI_rhgY{_n9bJ2vIFtczX zmFtqfSmV&>t?xM0qi$s*hG3&Oj0Ue5Kb1TSu)25hT_~|}LkHb4HQdBpUmfQJ?I_)c z@j*~-`*l$wsm1vWM1YI{!;Ojlg0-^nk6HKc5eDP&D#p2{(h^84`6!Z#2K2&WdT6i2 z7k%y6y0vN0IpT++K#Ap$GfmBzP!N(q4)GJEtq0YbUX%sRM!d)j1ZdV z;|myZ2O4Zh>dUNYD*0Wbo9sNLgqfH>Uo$@y_9M|R=%ht6_XB^IQigSl*#YFkb-4n@ zS>%cCe`xaFu$yEBono#-7-)1;O+7O^35Zv4;#paze@AKP-t%!kZm2c|1H;R#CPMS7l2UI>iC91Pj0BF# zj>`|`wyIh8b$-C(+9O$E$Z>W(Z^D+>CAQaXp}3h*GNaH?w!ao?5|>iKaolRnn+maJ z5quejQ$Hgep1V(D2bbp9qY1X&IuvtrPZ1ovBqh5T>7Ow4i1N}|1>p1TjcI0lieUg< zxX5Rlc^SuvPe%%~=@^ zDs7KS0A9b`GADc4T`Z-~Q5cG>Y5;5i2a|VXS zAG8!$In;Ce^yFp9_)|ZEK5)$4|0X{{EAt85T;)70PN;g( zgmrekXsy8tr2a=_jO;pzN(_^2uFYf(_*|m~Y*eEkN)4>hjJmsw9^^H*my_ba# zq@NzFR}7K$BO5N;YQq#*Zybvf+FVVn)~*?yRJCkeR=q6ivM!Y#)&o}1`8jJ%;V5=2 zp2wL%zC1yky4{L8Q`GtXnDnN7eyrOEmjURBHjDdN*Gk3;AiAKVd_Q6pk@JVL@95u& zFCh%FG3wqj;gq_lVM)`+Oo}Lb<4ji3K8SJPNitpByEEase&ply43d}Bu&!p&?rwzq zNj)Hlt|f@y{_*zp*G{FLG&lb6a<^cv(z85Qj=_Ylz_}M`RUg;8fSU<%VQC-<9>5<6 z_wkt1YjG-U;!XY-#&iK(gnV%QZNgOqiCEi45q>5;cny^b7j``csbj4H{gtx6uz8Ge zQIyy-!^hT@S(~=korjsa#(gm9zJ4XK?@#IvFM>C=^4dPJ?rMK_ul9S<^_^c$`21X5 zVtH^5-b!9!))b%KWd<`e{A7rB-i-K7?;TPn8)Y|*-|B>WgdLvzNL4shOtsphb|04W z9vCxS?Sy9ePJKMdD#*0moV~Lt4~==j4BiyWODH0T0Do;*R*H%`WL77t-(|`rBAGU; z+gie2wLs+B90}SBM^seJZa)UYwh4x=^M#4Tu0LV^8g43BkTYb9{gIivYfd;C?d@w4 zG?0k6JO~w4DDVRwZYf}X7gSi|qMVKeoOn4;7t|laHUH_+wMU*3jF|Y)eYE}!1l|sd#t@I79eQxe?V}Z4F zn?!0%Rh_Zdu&KKS&u_&NlHZy~K>+by!S z4O#wb$HAbzRDsUUk-e@>yL#!ns&X|2g3{p^`o2#Ox04C$uxXcD&O1*Htg;2Df@ysF zNRrMaR&Fw3+VWpy5UOCkA!873($G&Ww6ko!>I^2@lg`xQvjkyGnV3XO21#NLr? ziPi_D04zGc&!OQI$PWREw;kIfTjMqTcdywfBOvJV$&`gNdpu=kmleD=VnKa-fClj{h{pZ1Y;1$-KhbmN*^qBMh)UvIK2^^Tu9t5+;u$zGI{Az(D{Ie}_1$#8} zc2}~@Hw>kqxeUav))crzyLc46FvU=4?QQEZ@L8(v6+cgWnlt)%=aU~}EPcFY5gwcF zryM?3owKqXT7Qogi(l4HDO$g8?+<&)Y*s4OE#?f7lqY^#6Ds)@oQ%Z5?zb?3xgv@L zH2W+Y;Q&A^&0k;ZfgqAMp9rz;*Na(&kA={xw*yG8hl=7^M{hYL7m?-5cw>}3?!bc= z1IJNO;rFH*N5lAWh3~iySPek4^Xpb9s2pfboXU?Aqf))BXPZmk96JOb#BP0NVZV3c zSA*cs#we&?>igqUCD1O15~rKB=$RyPPfVbxuPirrmN6BWp{AZc-VIeCRx-prhA862 z^ww#ktP92m+TG<7pCpO{uj)OF=g!*@xTEH z<3o4E{l~@;zvT9Ma_Sp0@I3oOHIpcnA=t8df?g&6;}3E=(-_Lj4!8k98Yw9!9rEHo zqbW*v-sJhW0D^~!=C0CBRubvixIfbmG%aDag1l7M6;jPH?;X>ugAxImJI=<`o_imi zu#qPt`Y(3anWPH2Jj8A|A6Sde>c#%ps?3SG@(&YBOB|3K!4fI&a%#yHG@;vt)Ofb3 zh8Y;c5NI!av{^5R9FON&sU5A^L93g76SI4|-B4OjN*;&1&@Tpj%DM_jeZj?B$%Uf0ZH>P6mO(A@9GGO1=Y` z!JBazGxtogPqz9~fnV4|rp-Pj)kg#)y@+0}?chvjuicxo;2@DW|A9QZqUjBUJ6OD} zx8n?=obV*1qbSGK8QsCDUP}NF{W}j)yc9WS=UDSAjhiS+h%K(rSmnFFH(Ij=Y-@0o-_xe6In^^C$k$Pcy~f5wI83;BLY%AA(RAKWl@qf zWxi$lPdRVeLvJf!+i{)-`{66n8(eBKJJ8LW%IT)Y<9RQ$l8LD4I4>l=cxnExfQV(|=3_*nZ&>ZmkYWO97mK$Yox;9B(&(pBaG_oxsnpm|H zB{7Upl-gF6F=$6FNIA;776rW1i=Fl=*&p4bUmrKR+?PXF(vIn z`usAfs~dE~#$M2wi{vSO==AHyd>q){tEJ z@z#_Vv1@zAmB5P{I)I$$hD*S;r4wX+U5t@XN7NiaZ+F;kEQP{EFvp=S3$kGj2p1KZ z!A{v^^O7TFr#fAUElwM5)57hO0=;1P-RlCd0!}x%G-L2Bo^QnEx_i<0atv)?q;rg9 zLjl6x3`PPHO%}iDG8m4aj5m#Z`{C{-t+8&{G)w6`(dq_b0T$O`C>U!Yf{YW(eD{tL zH>`3X(4PuBitGKSblkzC7^T+gZI<1X=AOTVj)Q3XI^VDF<#5V#?dN94JAN@4P^LF} zI5A6QUC)iMGOfa|L<)Lcd1(=qBODO?&_?XNGzX@r2};_1JJ6%yfX5IYt%aiKG>{J4 z;l__e^n>s>d(TiI&f_V3uPus3;!&Ag^Ub^Q>AjQoc&}aaiez1;Ny;yq`?a@ex|izn z5RIEY!dUM=AHmXxk>clTdjvnnaB%&`i&Bcqnx2U8J<@I4fKL&&ddor0)ZCd^z;JLJ z`x~Am-@dGl-?Qj37&1!qG_qU#v!wch9g)b9Oy_D0DFxWa^c-8@_&Z&+a7T))lSL|? z8JDJi#ZmB0a0i)y?l-gUfR_wqH(napu}jbHHeacZ8T%72hj(w`$^p``Qx{O`qX7lw zfd&B~5cKF9V0PTsUhBW!-H!0YPuHFORY$+(aHZil%qRmCK3+cUBfQCG1%X6xk}8P^ zI3ot{^nswRH*f50Xg}rn=Y%mnNmy7rwdiou&`VVfQ>xQzP3rD{k^4N$6a5=qS###z^c%|3^T z*p-0$5u`*D0a3SX-KscMmJ=KmR#vkMbbBFFH)fEIYUb(Q0`Y9-(Lh+e!EIfFb|tc! zzak}%S)+0i{e)7`jApEIN;m3UBNrT?#X~4x1CS(==j8Vw#t-grabW3+25vC$wvBZ9 zWp$5)P`*ULAU8xsMbb8j zZ4}Jw`kTZmVO)t6B$d(1YP-)#Ko`O1cC@a3X3_^*8+G;aJAf$(SG zH}Cy|KzX-+fRn_C3A%@w%-_yqt4$tID92Qw%{J7#+AAN~yf0}7DVP?FZ8 zUC94%Eu#)&owAj9{?g(M2k;S=iyxMR|(JEs&oO@s8IMcM6d%p(iRj2D)lBv z`T^TGo0!X3zZ)a}4(>*4)I5&#oQJUu4<;RKqbLck&2+fjK7&BVKMPy_7JUmQJ$A;& z>n7Pl7quudK!>zuzzR-J5vF@BBkK8B|D6_f72^Q!JH#A%$DGo4(g35 z?(Z}xfy|-q|8_-=<0kE&u*Y%FJxyY(Ugjrp8fj0Z`=H*GHiS6Ui+M5;b7q_ZEVs{X z_4*h{ec8XLocKAcpRBXT^TD(6Rfda5?5R1@SjXzPx<>{H+Vo zKrnsULz~mXLUtnI@6uY0@sgJ}uUmaR#$9WH&HVB@c^57*49;<#di8Mnq)Zu4*Dvnh zzm?9dxe#$uNSlpe4Z4@EPPN9{HlPzR97AUgTphGIbdhyJaD(| zyYE~vi{qoCGn>6=O<{!2+oRZ!HnAFA#wUUJRnbUB1$NmuYY9z5S*i@uB26&iFA@y} z2HQzaCmGAc@&qybKqygfTYV1EF7FMH^S>v73QCRFXE-=);eaC*5GJbWq4X^Hd1AU9 za{75KXgs~98}^utJv%gr3_^*_egEs5nPtZe_=m^iF(m;ehL>3(H814&Q_|>_=_ki`DZ!W!3Bu!g`l9#w@6 zGt?GF3(C9dp3@9sk3$&$q`!xv0F^hHz{_@~DaEmmvck?t*rvbDk@%-tv3;|j&t~qd z$5IW=m2dl7C(|vJ5@uk0DdEE7YsoZA8o=WUIh%5%8k2bX>1C5iBvQtsYebqn!oE-c zaN+8u-Bmzy##pwNIKvo6?Zm(pIRjaw4%uhEnewajZkXP<=(#Nd|M8N4R2Es&^K+++ zP>v~BN8pqk7(qe^1v*CqBOiQFA2q|0X1#(3y6BJ=`OAqX?sJWrYejTk@eX?#P z7_n=6r+lm2++uEB>43S zM*m6;YP$3A^Q;0r3DzSs_;IL~a;pkGa1$#Bmt z4ZPFX$@>AcmVtcve^as#fjty zD**l?878(;Q5kZEP-SS*9;&ODG$-L%s%aI!fT|US*40#dqa_7n7H@rLs47^*1=J`C zBK=i#`~GoA1g)BWr~>2Xr4(HJK1wVbYsKb;*KMI_VSBL6DBWv2%4PpI_<5UL20bO2 zQ`RA4E$OauVZQP@|?t)QocVP2G!Iu>61B<0G>8@^5nMXX!W_t9`9HZ9m( zZ@Yognzt)94fr@4Cu^y@$IbjRx1nAGr`&hHK{Z5)1Q&MxD7fdfR4!L!dk&W;!WU&p zs`Im=O{I%C<0V;Z7auV$fagwdv5m#a3!w=NYjH;^;w_ZdWf^>KjoC9~a7E1sQ9a;bMS_-GximEmyE(15Z;EdGE?dIOSkcLW?nF5Qo2X3jEWWyQ%}~*j#f8Xx z_KV981}ja`dwLgi#WOaok?h`3GYg(Y=>@B1FY&7WxG?rY_+>FwNAFlq)per@=t*4WkIe)J7bi{?n5q~9C?`@wSdcc=VmLKc8XZLqejg~+|=7V3B1DSzS z90cdCZ%~jga1V06MN|*W$|bIBK!MGL#py2w^z1SI9)30r*k6-_NJbqdN$_@->@Tp; zMk6!(eMUprWSiQ|wTQe)m0HWStX%!ki= zSnerH(cNX?mn~G4%55cP)7_8C3AW^hiPzQ<=6TE_Godm&m_OH|Kl-c}n;yCD4Z~B2 zR|uZMEVYRqiB06&(XzR?PNd0y7w6&qwl`K8E4@}y|2{`h)5}D?n5D;3%BK=C{7GW` zti2N&ObKanO|v4D(*_J-855Gx32CZ z{LkstXljol1TJ$X*ujlI-p%0-tM5_v`z43UlY%c$caslSZkd!*7A>oRM3+#nXaS!RkJ5OLi3R6 zj{HORUcljGO}3F6;o5j$zsmD;_}*`R*Cb)E5f{g&$a16N%lf zj23Y3mShQ)3CGMSeHhlXNq73i8@4$qySq&kFGa}$rp1)^A|j@Y1K)li1jHIQ4A{u(W6%Mo%Gled+TLNwNRX;!O8ywJ zrraK{*|*|ax~H;1rFfthp{t$Fh6M}p+mLk(ACcIcoBVcO4tbI(;^Ox?%$J~ zn^tPi_*G?h_*Ekm#B~70tnr71>}S-!T?8O-yT>{pBu)HK)R#t8&Ra|AQTV{f#ZY&z zaXj67>Xmm>FK|n5n2wdTW9QXK&P(eK0so0m=J7pIkG&v2Gz`T-C<8eo1};}y{)sJ& z!1ubB#Jp~M^hfl=FcVk9kZQ=^EVm4*zvh{(Ru?gYk0~Y^daj+4P2#woWMv&HqxQ&} zB#MypS=vaLd(hL|f92z!)9_gpiz|Q87XSNCR8n-qDVd|^1*tG6^#j(+SxNr+E-5I= zD1u#dh=|0fXFE2u(OeshnB<)V7RBYplSrhP9+CKV>l_t~wBR;=Zk{s$PVNswf30+p z|Gww6b%Kr&emwaKVMXm*?Q?K;?Jl8OR9B20$CgUvTie16D^rm%trtIW6Ssa+|3vD6 z{OS@6S7I-$gO~l5d{5(M=%pW`f1h9saUW#vzGg@OrgOsqxX=4iHOTr588zy1ysC}A5W=)l31U>_IU(Da~e^^rkI-xSj?J(p);_N_lD0tdnC1= zwAJz{m>`x3f@8!*jkbGo+vCQ_brf^?PhXPU)S|4(tAZmuA|!`{&PJ9&oN2 z_d*ij0gP4q-2%TA!O++DWbDf(Y(H_STGY^ASz048xpPZSh>|+a7Iu8~s>^AMY_jhe z8bef_Ofub4KDh4xB9NE#aOdOJ1zB}x-j3>!EeXN(%ma(ag!4ow;DHl3?D$5iL86kMO;;X%#RlD5sd34HB4gbDnTw9xFW` zNMwv?Z@5}y+=7yvN|vj+AeW)z94Onvq?!d92%sOwSk>L`Bba(~#zGil9>k5{>^zV5?62W5@{)(tYT8p7kZ8oXc+e0p z-+kj)D!3*#f$Ua=ZRe3uVsDeV7saf@9?Juu)^7?~f!m>-8_6XQv=^1q>I|0G_Tw9L z__^}g@@_s5fT(tZO|hM@qmSB7-Z`rGcHJLGR-wyoCN0Az4@p#HJdhqHVZ%{w>8zTW zJ~U1&zY)1(h;YdipWQ@r9U=)9THmlwHjJI(WizGW`J;#2sxFuO5KWoTWLEz3^y5Qk z>Hh&iK)%1l2kTk=T=*zz;V_uP<4U|Zrv$Qkj2_sSO-YfO$SHOZ5h8ZXEvLY+CelGN z36yn5!;djA!+@;5uR<3LL)%9P;$DT6{& zZxoXl3XvMj3s}q~2P(WBTVT+|bSohl>sgQ=4B0eD;M~q|_ykbCI=ZJr=e-bw!rS1F z`;fx=jrnZZ?tWc5?E-bkisZkq-=aLkvADN=Jm(BT3xu%j%i4t*XJc4US0Jmo6#O0& zLpl&dYVF&yZnKWe`*aMt+JZO4;XA#?}^#_YClj>{ojpk>F{k~96 zf9eHiF1-=dnRkL;VN5?<(E$A#HQZ@o*tLjS)}`+mX*Rum3fbJ&i_4j;NyR68U>R zlJ<8cVk=?*EQo65OFj&D5oJ;eDcYkSP1AH+8BKwU&R_$W%zFxWsbvmdq|7;jnDW*Lu!1WiJZe(ofXQ z94{J!V0w2xZs%-SrJbedSkxm;%6`h|pJQA{O}9h(K;AlCvsVH_$i^&7p1M9;3R7zD za$=N^Q*VhDAPLjGyYx(%)#D<~b{bS=pd@$F3CDqh9-&?Rs=(pN831A#X6~d>-6F{x z7<%ca$3uuSW$V*$d`E|GZhjrg;7&9in?-2Uelo=rxA&RQp91$D2-YK9XlGlc`7^BS zCyi}&uGpcIMQ2DY$8G(Vb6V+eTfNIS%%OcOQ9EWBFl-(~AqEa=g`gSFk2lydLcmsX zy0caKSKeU6`q8W}t^fwzf(rvXTX2B2Pl%Zo5RJQXdn3{Z+3ivizMiZv;MoR$5U^V& z12h5uXCxD-a|sOU?!+l4{V3F9TQpDqz|x7c!6E27?9hD7O>B@S{7>w1&!p&G7~SpI zuja!-Y~-vh-~XOBC6ShPqS2K7-AlX1)*9@R?7y_YMcdlWbeADO@7Kia z-K7G_*-&_65?p!`N*|Ha>|q@VHNtC_$8n4sx<2QeGw`~RA)rfypm&00tX_N8WD7WO zV5Vl!hDtJdJH$V4*g(6RV#zrV1&@at%*XJCn0Ps)cP|MYnpnc6QyS)OngUMzhI=Hb za+Q!*-a7_zyJXuth{4qlwh&lN0~SwK`R}8RxuG{8ZU2vWBcPI_;j*CP7jme|%@~#t zOHsO?VRLKLHGkoqylljwb2v&g;;NtBeS(=u;fY~5mF@10@sJW-AuA67ChR#BWP$_B z&2diM!1QFf;Oj3;xKITNl3VV&fnAwn3yz_y`8a{&9`JTiTBHWK&WGWe1n>dI>x}1C zo)6J=_CpiV@+nJ|Unn{-Z;t=NJsN&$RPHCmAhV!Jk+P(WljkGFos*ZWdP2ctgyUIo zz^*Pw{RmIfVq?wHJbC^kJ@WpYACfyQ{RU&+!R8bMaPns=YmiI}D~^2CXk?-()YnpV z9vTGW6lyqa0-&-VoFL#u`XbX7Q8qkFBH8(Q)=Y8TJr6(^**t5I0h~kH`s2050L#n( z01?HHIT}~n+(0#V*NB|}Hj*W0o^JU)d)H4Z5JS~zwM4cv4YMDj(gP#a+pkEnG$7VA z{l@AED7i4?qDcrE%GqcjGcS@pw7Sc9;*I3%hN{F+&9Yz@maX?|soSoE0`>CA_$Zb$ zdq5QZ)+87uYL097+fMvavrdH;0ub z?&b@CmeY^F-M>S*{ZnT%^t7&plXlF2|3#`tMSB^d0@Jl{Q^QAsS}!b~LiC+Ly-bgK zni7%~wao1+Kqr)U%^-#L?=n?y&0?GtK7To%u-53O5*m{ zf;2}J?eR-{NrnQbx$DM3TlO7WEg9b`w&1raR~a}bN!?`P6emlKi99)2WI@cRq)rv? z;CicES$@gj0M{m*7Ow|ilo<`Ws}K)v1Z2lxE?V2(*useGd>j-1bOMA%s3H{v;65j! zEff8Axfon%6D3a1LeOR#ZG24D42uD~vJ+;XX+nH&NLyQ;;UXTkO7 zd3TW@9LgPJ;|&BymJ;@!Yz@U(GSqO(vBf@`^|U0E`VDMaP>M{(=<9{)%m8XlI%L!r z!d>`uw^M>RZ1}AF^vjTCvu?rBGtk&b3?F%xNIeILy~!9W5vuTLMiiM1NImavQ z$yoGnwWL1~q_|!sKHM~-C(F8Q(X{~2+5oMMeH+J&IbjOT?7xZS z@$JB%gJP4kwZkeoscYFO{u8yvk}S~*@evtP7WC$c>)iNjUleUUb8;Y_ zdTjT|Q)xBA5TB7Q&k%1Byvp>*n~e!kxbUJ>e^{z#nocl65Fv{dO?`>gd94N-KK@H7yH|NCH>b4k}Y2bOh zSKJ$dCy$lTxpE7_%tqEv;0iU^JZKm}1>bZ>Nqt70vuO7pca?Ti)kEZr?$M#22<00OiOK{rtJ{Ic6oJSCw3kP(_n>IlM2>}F*F z^pTdRv$G%(U3=?IXrmU`MEu%DgNe_PjDhuJk0=l7d05|VS=^PS)KwiQysdG+xcaqZArCd6~ z%|*^Ox#hLOnbeOZ3UtYN*YoQ?eh^&R*=y0%?f(7%8?c+ZQyMt3g@=|@DJ%y*cfecF zIqZqMGkuiiQ!N!H=SKlYk%eQziAU;C#Kp!O7AMjwdnzpfan3b4(N2Iebgd5~GZEtg z>dm&Z1ml}hIYpo*DKxoKf6&YAC{Z#F4si!vHF~|WkJ;{`U5x(Ai*iFKBccr@3Bybo=!*Gg)bo-0a=rvCU8+J*YxfPQ- zU5WkX4bvh1dmRIC7uc$WiL1QR^3-n5_~9G=nrz+bVR}fqbF7GOL`AnH2sYqezVUTc zCD&C9>Hn_kJ8U@MhkDFmreK47Z;~;URd|M^OhMA%*_F&&;H%F@QXUl!Wt`P&cysJC2weEG7}^Xpnj0CVURckE*OL24J20G(5MRA6^^p?MGQ zcIH3qMXVzg7ChhK)`&wx5Bkc*`#Mty$#EWLAig-x2HsTs=|(#H;HIk(G6{c-ussk& z9oM^&#ovY&q+YgH-_JBpBL}pYhyt0b>PJk%HR61$JaG-cN9Jdp zcUk%cgyAmi!!O#2KHfs4Jrea+uWMN4$$fD&ez!kem*+sAd3J33NncY|+5ZLnQUAc3 zMnbn(xO^8>H&Wd7t#*Ff-ZhgFK{aO9oo1|I&ejuPxd;>&Xw=Yf>za`X;e3tdasm1V zs?wdJwbR}DP1Q>h-d~Ai*bH_GTc~*6)K`){RQ;+Mig01FANs^hFQQAj(*swj*)aW; zP)RWv4+f*&@t(N5>V(lx0}Qmje|M>f&RyXpqSShwY5@JP$)$G~zdvc?%%r&@>t3F& zvct$=T{*ygmUkP$>5$t&y|fQKM421YC`!i+0+soMsktq=UIb$Cyi0*(T_@s3dGvVw zp~tlNl*cq=X|cqqyvN7cMMA$hYHTF;46it>DLNa`xU*(b3ZfRq=B5OgIwIksFA9A$ z9#qySLv0n7{s6$R4q-m-@4Ln7sb7T&xO9}{Y7-T7)yv2|aQESZ_EJfKcwI}snya|T zCrebdWaju457%^hY+>&(@=>mU?Mj2t^Hb6f;xjtlI`$)3#2MyNs&zc{0-N$Py8U!E z=HNxdhiR+3p<;?t&Brg3>2i0z6P)&H%_R>CVa*!cISIir{CJM`RCeV}uX@N=cXzOW z0k{8Dr+o`AJj>XqQGDssTuzy>yoY-HR80r1`8uVZ71XYThlG@y+#{D?Y z`8 zHxD#&MaU0LQ0_J-1*4}^AZ5MV4z@(_e=8HLMqxIA;g{W`@unc28Vu}&j%UT>XeK&uxO+yR%Lk-$a*{bM(A zfLnkZ9ro(oARRx@K*=H_oySi2vYyL7V}(EG;+rEzjDnm5WFrtXZm-!4V~|kq{Zpz= zEbT*VrdJCQw>6@4DGc)TvW7@;&8h#3cCJ-FQeQTzn2|ng`RobK#}K>|rfR%9(=};E zlU)>6A`upM7i?I~&+YJP*w|cM1+$OU3;q2>s1LYoJL-bFzQ`B8HuFd7@I(|glws3A zKc(blm9AKJ=VBHv2F;8OfII1ZrV>y{Aab5V5VazttD{O|P#9wNeoj3hU>)V|Yd7WG zAt!2R@5mh*tGGAW`Zu?kwq)FU4kq5PWfZ-0wYjq85_SEY!v_#zm6k8^70Cm7H=lq{ zx@U>m!snHHp2AHEjdm`ezFFr|oW0Nmt3Ek5#8X7Eb?( zcM{4HLn835kkHTEZhMGppl)ML=*NudCbmyYTIM1pY%!cwnkq=0cREYDd0O8FIVdoP zQe4aykICW-IVr^Cal4u%#sKCQ-MYfq+XxDPvR2qrbUi0R7Qh)e9y%q%>L-m9KljnO z&WP52M*?wgr#!wnKV=~pH&f{7#5SN5EZVi!RnC5AWGQn?5TUZxjhkkF9{^4X1v{Gr9`$v#M zN|%1KN|?Trf7n%Lucgr+klWo{ZFtKF8od%9MA5`arc-qjfKd<_uUPOR25|+bPe7Y0 zQ3iP(x(_PfqQ6S>=O3 z?rbkDWGUNvA#g!_TQ#RDWzjT7OM0ju{Jwy{u~kPt|%6tXj%tJ|Iifx z0j~Sr16_6OH;(DKqkcRgxW@&~`Awvs0*dB-L^TPpe<_wob->j?CI=$J2TS3Vn{$A| z1831)2Fa+EjEAB5tM&oNVI52pn`OL`nVk%syF}*CA05FhR9ZJ?*aup&Vb#b6^+4Sz zJ`qufP3Apuv?#1o7rgxw>ZecE>)O9WgEnV-SxCN=bSV5~JUsKjZ%SWef<7iaZ=_C- z6Nh$%{11@Y$P$)fa=ufFAPWaVi1O<&hp5W?R(&vwWaNYi!JlLPdro~XXaJTu2aJ`x zp&D{JKa`sTk|-uqvKNehh_^({xf2toF41@p4ayLZuG;?(vX|lt%_crryT}kGLhp!= zHMbD|^GA+TW46Xu)86UU* zrT$T$c3Oi4OIw5WGU}4`sG^Z)B0)-z51eO<{PZ6l&#C^;k1t1Ul`?M^HrjS{VOsj! z0NK8{KU{j)Ya4o$#fNiS(ouKZ4M=|j$}N`FpK@Q5@@0_VtD5tH%(1Qs$BQJmaNnS%!uVUw`9!7J1AqV&2*pcg zV?<;3rG3FNwOV7MS4rAeq;H?C&7X+pMJcP~+&ne!X&QmFn>j#pigJmP*XFeNaxW^C z?j<>CH+=xJ{4$qSMKbS!|XhMtDzz#8}*IA0_v>D~yZf2P6u>Ncb&-Z7vR zrZL&=6F7x0zs6zBV=$!fP-33`$=2(MxyDys3Rw`J$!)_2k_x^Y3T7C-OSAhcq zFLUDj+#U$@t)_bH`0V-7@emFnnJxv=MfdS=)f}bGY&RaQ-Nv{_I_ih}keeC)<;^2A z;}P0?w#+OMVfgk)H+4p!nx-8nu2q#b0}+YJqQBb7vzY6DHB1ZBT~}rh3zv~Im5rhg z8uLPe=`tSIivq)30(q^kcIr+}Q_;a`7@8kVl0KYBQRB^fNqI+;q4LgneE=Q%(Ulot zTpBhn+2})HBz_4W{QphuBLb8AetXFyY~wZG^b55y#>Z=K_n5U#4!d&sykwov36goX z5me`%R~$9?g1^jI({+t;fjKsc`(YEYk~O8lUM1g4QgFBkx(~Q>)a-kQ8K!gHuCzLM zzdJt|%VnJ{;$)&m7Rs(mwGG)y=K(eSlgoLI%=-50=(;Hmhd+bjQMXM-QTo1s_ms}L zE*>ED`l&B^#n#HFaa~jDVA5wC!Rg_ct_J|z*qonYk_oG>e~X;`09v+p2XcL#{xk=8 zGR_5EHT25j+dO4>-E)-nu|=$51Y&f-|C}RPZDkn|2JgOiF{N0v%fB5kV3e8&3xAm90FVxRwt`Soq>tf z3PD=Cmu1jEiQ#{E*mh5nO5+IB((bloT9y5udww z9JcU}kxUf+Npuu~HaQk?E>iR{D8v>v^&l}UI=NZ5TISOEJSWB07B{!iUk~RaA4i{C z4zw?T{C5VL(+oKOA~&%jU6GjOZiF&(c-Mp}6yB$?(@g^asO@grHEHv$-y_|~nNKZT zGiNS1L~;Fqpre9@jOcHs;#`q7VVb`Fdiczip@h*_`rrJp$1iFk?jPscRpn8Gu#-(c zsg+TjaBfs6o&Z#69;th9yC}pU^RyI_pX3y;zF>{$sjVI9>BoiDWVbQb^LAk6@FDzF zcQCr2wFB~^$W~Hck>gd5ukNfv(<9*38Qtka3bA_Tqc-RfL0B1JD&3+b3uX9cIH}KjKW4V$ytkNIvDA^+%xZ!b z4Nru;f_6GVr15NbcrU#KE5dI#!)2Sv2_99Do(&WR>Y4ijydTrC#6}J_T)UIv-8tPl z9-zH+Ny5a^OjdIZ0fm{Re*$tWn()zd4U?hnsS`)zmfkJ2`s!4y6?qzwpcexbu_zur7PRV??PllXZ(&aa`%T-B*B z3h*I*@DU|Oo`%ye4n0-&mBHD)92*sP_EYN4erQ)<#)R)ShPx>1`?Qs0bqUi7HI&ETc0IUbL5gxzxYBmuZF{6wVe`cUspWj~9< zn`|$dj!PL*zLaC7iKX${>ZAD`HKR_vurzpD?R@s)_5`=Yi<_(*7Uw&I=HjS>NtK6_ z=KuE8q-0{EHEuj#b`Xf2vJ_U`oVQ)e#wg2a$lNic?G``=14EQyh%CPIufGdSd9ON@ zyq|DicGQ*R6N}MKDA*jhY%z@d(30O5iz*wFENRIp5>GcflmvZ)O4I0{$T~%pH_Cvy z@c6t1owj|Z%XB&FYegX+@o=l&S99z5u{VIiUV|{6HKFD=JIAE2+Py)53!4I0(~5Ew zz!?R$rdO29H8^0*5$7*r=E{mX(gXXaElIzLhk3gRk&>}#g=h~^qIeP2c-r__^K(== zGVM`lb3jM&%sL#;WHxEbyIh4tZ2eyK61zKDHiZUPN1zKhHh_9diE!-SN{=JaLcdXE z;+G<_i=YrY^=@u#wMCeHco=ut`0a-ZiM)`-iweSTi9qjgmA`-dCL&GLghfC%%6_EE zUz>)F+b34EvtD3YIc7(&=NKT}miFEbI#r*{%`N1~aryGiEx;W)h8yg8Fa4~|)8oDW z16mW@6Ns8r$y2iLaY$1*!4}s2*?bI8Hna=qVXDZ0A;x=0BF?kC$)HRa>fcn7_O6mw zd!15G?1UdaY$Hu?D2aosM2&B4C+49!Va)-6R3elCgZ-=`GOXWaffEzt_NI5)8JB4aBN$h zem=U2ZNi+T@r!+6l_aaa(Hq`nhNuPvvHiqRG<=5TV6yM*z6t-%G#>{5W#wZO_ArB_ zU>;1Lg~@&|pFee|6knHLsQ)9r%cv5= zk%LckU)rj*`MHYJd!lQwqV~}#8XRiqM+7?^JE`@t;Y|i%QlS&&>ckD&>gvMt>;0&g z5$;59LKGL1JKXVX_1MSx&ezCvi_-z?KzXv-^7$QD;m#xjG8@18Lx~#k6Wj_q8||+W zv@_%{B>mj6#fF367@(^NXL6-G%!GS!rl5!C6*b9iIi9Uw9S~6RwXzcZFzw7g+1jSf za*t}jylKpx{;e@3$WwE65_rKYpgpn`QrG(3DibHr{N&1sqp#P<_rv9Ri{{Ft0Qe>I zpuC3iZ|(|~-&yQ_(9fhU>+>I`xDqVNo;bB3I_q2M-=od?0Sr)YDii?ywpD&xr|M6# zV$7G3Qr@{{r$K);g#zS0@0^6N`gPm~Vtp!BeX9m(eD_!Q>JqGMl+G`nQr(4l{(R% zMB5EF9M+eIQ3wvwoBW-y9?Ht}K4ty-WfoLK_hAJ(AA>L^V|bzZjUCoG_onD8%!c?c z%=n`q$m|bEf8&N`g7f-O8zVK!ER`ogMpJir({|QpPcT~*`8`RQo^hB{4$;RNtx`(` z=*LX=2-W`bCs3=Tj2huelqm%dIi`I{J{`TFnP8bk;CbmNEFTD3LR$G0xSuat12v1A zu!L7JmjIo29!=Q`-*_qod`?_ z1J6ww!>o`*)O9MWqR>JCCMQ1PbjD(8Q8beW7B*s8Ve@0%hR}hc5vE#v7 z7ym;&aO}sRuP#tp{)!XFe)Bv=AT-Y*U=rVucpD^Z6?QDA0)nn4+d<-Uk)4H7?xXpk zIbixyC|%|p$o8?NFM(z(lNqO;$lgayPrq00l6T75#1vgVKtUdW_~)&s)1fMWG7Xk1 zctgbQ^O)zm72kL=xUFWRM|A!d?8 zPqrtaPce43)*^@@Yy!4rQkp?I?>M3WjeS#C!rg)&nMH^Tl`%z13_e=p`&fgEn8$$9 zi^{-g)dXTt3Yc?Dl#(te(P6B7h{XD+$au&rL`Nl*-JU#3@Vnq7w^$^3$N)fZSNOW6 zC~VTkTp%V)07WcHkvaiOc|`U>`3hm2r}&J+za;XTr5k{$kiuzglli_`lM|ZYNHD_q=N7LXA~>a zq#6Ma6^ZvB{?qKy2U@;eoAU#wI8}7w0{AT!bz}+05eQV(r z0EH~jsM=!7$V4f}g$A4Oh7FRpsa1qB34cpZfp{yCV~V_L--fFI4d~8ZZl(icYf=T#^D^HEN1F|D&3EQt$dJa5#EO1p|2N<_W_OmXEChs2!Er=VR?dubC4Ss|&ZD3Ip-6SW&S2rSfuQ~^-WhzG?)`jry$h8?oD zk^-}Xg*sS~IQBVlD9BG)1x(khYbyyut|y_e`WWcI*z(%m@FBX0XN1NMrYAeS^turfgEQnue28geO&=Co9uN{Qy00 z{zsJevj^m7h9@{<3UBYa@ql6H@F!4wEKP77x36~m7xB*IMa1>_P+Zb%0;(uz`C>w1 zaNtQ($m@LWP>nL&W@A5Io9)V%HMK%0WeQKoHHgkbi&gnZKN^`njScj(~|9j5zaZCs{q!E%Lxd433 zH;3hU2IOIY6A~XB;s+_-%8q~JJxTPI?%Aga6Fj1axDjEN?w@=75M^W#QY;UbOrO)y zY6=jIxyKlliREW13cWESkSj$rJx|{d7`?o%oA_J(foI$x^l{YX3eGeuxaD>o6V&a2 zBq5n1$jG}wU))ctqaHsuLQXBg02RmA&c0W5pX(%ZlW<1qF^<1fT3`H@H`R{ZlgyM4 zh55+0gY9Bo?6D{W584z*TP&hxas!KlScQ}X3TW++4h~2|x@ju&)vNR_P|7YBm$D~b z_~~`AVJ;~Jcz)m0$K-u;E#>&%9nUQ~a@{Kl(j+?+;%X1r4cc@T*aw?@#;PMvm$ zMu2r2S1RhN#DHilgIe%#J7Kch|HHh7Y4p~$e-$G>(HjdbBWbtrG9PHs{5C`Er5_~41ZJr045jDkrkvi5ud3w`_BzjJ@ z#O03@(2bpRGe?Wln4s@t?7#O~R;D#yBW%I=)EmMBmcEMA(rORO^}+QkH@-lz22tZH z5P)2_l3u9S;yfa;ch%-D%{&{uQj!mfYxv$+a%UNACXhTYYyAE`Fy6>U7Ou+mlq-!7 zL|E}qr3bfge>S(=rx-VO_7J-iU<3@<5T;`T8o1A)ZY8!vC^cyxiJr`}V5;m>QPPE# z0|9fZF@hhBnExATAVb}9CtvB&b_Sk^SC_12NTIB63Z^6XWv|!gAy=zvXeZ!2D?V3Z zo4<YMA1xAkqNpq&^oD;Q zb^X}9$h2=%uC}|Ky{AiY3kXt`t;Km>Q3wszX)@eqkNK#t>kMj z+sMW>MS+P3L>gsE-Mv;f6Q7U!0mUU-)Phi)5`i9uKM^mm5J0Z}`aHId+nGotc?)p1 z-T(H_)w^V1l>4cBCF_!m53NW5Ku${T)h@dPBx*X~>cu}zi^^*Ps5i!j7ejNGT_K~{ z>zaVfQM<*xjw{2iOMzxadnE*gi!d`J2H&X1wygX7Ab^IP8Z40av{JD3KUHB?%?)M$ z<9$E1x!E>U@Zt#+E87SPD(TW0T;?cl&M zD>RB%2T;BW$K-01$_8EkDA%O7ym+q#>!_k{b+Y3My`aSc+`jXrd)_pOYU!xipb%2t zx_Lo!lZwkV3+@6^M8dtzfT!L_$T@-CZZBj-yNV!E5?uCg_T;51etL+itx4wDr+)YjFrig+RrkCy$5Z8AwjozO) zt!G+)xdFC*4yU~DSIQ>(+nD$4cjldG$I&W9JQJ#RaIdc#9KZ4>aYy-7>_s|yw11`{ z7R=S`=ASpOq8Z{cm%nC$sDR{ZT_r4NM29Tm3bx%jE)9^|zy&||m|);hc3K%w7R@NW zD!KH&Z%j@MruGM%BU|_uYPp&e$xH7hBu&~g67ewP6ioWG|1hEtt7Z<)P%JxU+OGMS zNV)t~O+Ju;174_JCoVu&@+8z8cKD}lSNl&J*9tKSk3gYWoA|f>wtry22)#6wx_ZyFzOnjX2*ehB>-0NO-U1~V=+s+5D>;#GXaUs!^p04w$UxI9DMM7;YEyE*c_u8)DytT z-k0EM25;REf@3D`H(t@zRnUKvKxM>$!;+zZel8}SnU#?d9yz0J1`X=7vU24Q9Ziu4 zj1@Pgv{MZVM>;*doV>sEQIhuG3&UHe*Xzxtr6j;AXjJOjc5Gk z*^Cje;~}a(lt;tL090k1-rszl7Y9ko3@R_rfB!M-H<;O|`6OAqy$_<@$mk0UA6|&a zW5;#s1<6}))p~+tEbTGJ2W4s$xCnH_Obz@b;Q5%ys{L7{VahgxXvHF&DJD@HYQhUt zB>Hj(VqQnmIAi3ITVK%U&70jeEs3TxK8!Z7U*Qb}96{RayX}4qar`fHQ6{a!i*hQz z^K36f_q0h`BxtB@)+g6Y;%sL9kpVZVD#qR2{ZK(ZUBgq3Ub9;w=p~V6YWj}DzL==o zU%YDym?ctqeV>=J9H&8_6!wC4E|&C8;K3an)z>~agirJ?9R#@@?S z9Ip(dxR_=A|n2=EvA|HYO<@m_MMmm*2yVPlU(XN`m7Y+X*eWnzAf(H!8Yc(0f)P$B(L~ zFc#UzwXOQm66drxpqg=cf4BAKUtqWuu=ux+K!o4yF7}gofDK8FS6L*g%21i-GTZ9V zS8z6|x??%|=2F;h z-1z2s{Y1Mm%iXZ09F#OnK1k@PqpQ)gf%?)qU~L%`zsS|*xguh|`MUH3&rPooB#n(u@w$-h>57UU?|9o>o~+yj_$Umj&o zd&fUE!nJy(0zI8i6;&~hvQi;sn1ovR>L}}712;RzY7B5aPkpwV78(Ox0goP%&Kc#r z=)1#?0#59h=!f333I~%&zruxb7OxAXERUn*46O-Mg|@i87+f`H!9S}OeA5JoljAMa zR%OWVSfXU>aC7)Vl)$W64zSzULJ>EF=U2l2ve2aL2|G3c#_`8eJ_rv@^Y7G+Vh$q9 zMtBqtWnS)kE`o-KwUfMZT`L?230B7DLK3|uE6Y|Rs8Nu+ezSO44ZiA{kWxo1$;$;N zt=eZm;v?L7vddtfGB_iWuX}w}Z&>0T*9O0pB0BL#rQI(Y!o!=R&|Gx-EBrJ6#Swm0 z3hFkBQx~D2kl7;=H0j+^Mrz5oUPu5W2+4RQFTd|_kAl9On#{eTrJ5y!;*3dT4)MEW zr@$}MdAl5=#lr250LmEXcy5o)LBXs2kWt&&k8ERu?xN2sB?s5SKS5@^S7t25Ee?!6 zv+SY%sSy{Z@p2A{xA^6MygdWii-yQHINN!|o*|#zbV8!Mt@Kvx`6u*!ZnjK%To;ofh$RqVrP@3)gyeBq2Ss6)p7=H6)7aU7bJ>}d>>J@cft82C;wo;&dmUt(?j!_d#dJtu8-pIaCP4Wa<`qW{Ck+(# z)HkBXGF=ECa$QyWJO2TW(u$J*re#JJT?(Qd+m75jG5-UuytF{#6$2#9BG6?-wwp^w z0ATsTm{&5Jm@gH5I&Cs%@Xli(@rNTCWgdf1ooMMJ7NYbTK;-IkxrX6?r<5l+O{F!4 zszdmQvPGPhX?|1(#ot!8dF=-*rw6FYxwiJWO}K~>8pkdrfQM2?i{KDrkW0@iUuIK} zLC*1uWAjDp)6We%9Sk_Ckq)8(y&P4QCtD& z!T!pW=hbtR^3usNwvBou%@!+FOLldoqYq>%^DQ>46NV)G(t9XY2HWi!!*&OLsc3cB z#yLOmg!uCEsoaEJWcVVUB`I@AXj^)A7e z8wle(Qc{^X*WdW&gU;MQ65|Uq?J|`fjb=!t0Mwp2vT)l3-ur!0tL|Ns#M{e1^C##o zd-+goXyasG?=aq^k&P6#=1ys~u^=?}jO( zbP{VI_o;gDra)K*{NpCAnJ{3ik0oR?rj5HNM*o(1@G6{asVTPp*3Iu}sj!Xyu5#!F z0-lS&`S!TaR0|8}+9S`Y@V_zb3}etJ#fjebK^}$^XPo`z!eV$tpa){%4xHG;fP!T= z5fZRCmg}X>HVe_?4Q`JWuz9P-;2GWZ!ljYMf8*?k15V?nbsc%b5@chXAXTDRIt8GX zt+EC4g~UJM@~&kny3bxqegAf7OKc-`^PXw&^}eQiq~9UMFb`_fGxHK|zjFU`8PHZt>>MB_^__G9Qp{@2k))Xbe_(iGgQe#(z;VRC9T_I~!bXrj_ z6uca5ggM~tRAi+C%3wbP{{_wzlbz)}8ZkupClGy0lAu|r3pzwPARW0OEAXOzM6AeD zQ8T)zFkr2{<$xh5lcUC0U;9Y|AcFM9&xfwWk{)~vn786VT41ZXNffuk!%i~yZ_g-L zNbj^HkZ0C5!g#ZwHpR`6hpKm7_o2+|0lH@g+YupKf@4jH(F1JZurxwoe^$@}Ahz&q zjQFL?2Rh^?*X?s0H-9TGG%PHP&iojNAbaj<$WpG48U;MW;s|)v0)?!K$bfEk_@&xd zg2}NtUMBPrqjMD#f@hXU-QNE^jW&=XHXRoW^-kGw>Jair;U)DcVi+fV@_?kzvI#c_ z09ReZ-0S7G^P0>ip?~tezyv!t5}UZ8c%ZzNDBh~tED;@{@foL{rpDbLS(rt!B0hY_ zd}MEQRZTtVLD@w<^w7C zpw^WPpudqi9_N>I(Z1B|(@((s8beo>)3U(0y|#1Z4P)Gd_!PE|&!R>D`@LPv8+&mR zB5Zd(gdd-c^O2&%gO&!!jU7Y9B~(9zeeKPk*n>TOTcUIrV4E%5j?fI-o5CSofDbI1 zo3iWKW24mF!)f&3v5^az2wN|}NBrM03<=8N-!~D55O2)Z1m>k!7JLNEyLZn+_tITk zn9(JWOpnF^4oiDAN5G2wJ3{(Xml^=480Ma_yulS$GU%Iaay^{GYcEBN{m7@1$&ELY zKhOZbWjVxy-b^cU9m^Lu0mK2LJ_*fdYDI;GXKVQ(zWBkH3u9W=ZrQ zr#$1L%ZXRw6(iQI8TAvV3>aa?8YkEr(`q8-BSVQri+{8e3=g<8Yrr*93j-o~p{!7P z4Q2E5y1?p|TJ`2?!%Qc~8Y;G56b-9E6>0F7~7Y zEx-MHF#h6bcDDA3fh}z^`cA!5TUDw<)qLw$v47<| zpl!_f+a&gG^S?6|2;#iwe;cjrb^~qm&L+z{FB;{EaG-;$BNmfDJu1oLgjO>LX|)H$ zL0@Gfby?X3AGlky?{aAyeM?> zCI2uS;sds3sMa0#)yd8NpA3wX8|vNPilI@;jjXYz>2^%*V}e{PRn=O<@1#q10%-~c zt-^K!N1g6j^V?m1xJ2F#)I63rB1br0eByBpAB5q6tySF{M(-%7dovVwv9nGuip>_Cd; zHdbMwH6K%gU*T?+Q#d!}^x{{L=}vTu8qRe_c_LQ3)j3ELo2)atss9zBP=xnAbWr28 zn6*A_mf>tI6!F80Mf&E-W|}6nA1p7ziCcpLBGAB z?L=uqEggtWnkNrM%WvmM|JnqiyPk@0IaoIMD;?0wqk<6%K+^--QjoeSket=Rq8%-C z%W&ShOei}zs;GJZ#KX;=YDZ{^1P(vEt6zvRFLj5Z9&WhB-iVq11oN}&%d3e)c-X5K zY=DtiF^2}#Obf5LoojMPQ!x6l*qWzTOa|`%cw*#@Mvp z&;HCrn$qz|SR(_l5SeLa{g71G@wY4&@@p&1OEUH}sZ${}eOkx*o~hyZ#h3GIr`#*=h_ zzHcFOTJ$8LiW5>R0Eb?e`><&&b3h40t!E%z_A_JtXp9pKq;}TXIMQR8-IRz6ZogV0*q5tWvW5~7fni5_5dvNptGTcu!tgf~ zKs_(JnjAFj-)dN+oL2R{_IQ$1n!VWMwywolB96P*NW007X-TJM+OPx6tPhv~>`xT| zq(p~lK2w$X@nas(Io$ds>dUZP2r7GLb{o5y-W?6r9+?vPBx%Kth+;(s=%v9L85?Bj zLL|ob01pW5BnO(W-YzK6$n&}U2-A;sFUd?&OR@%lt=%+nCaZ+NQpT!NV}&F_#lEra zxU+G;6~-mc7>z!-!}L7F-M&9ZHSNeE9&!jB+gbY8yFl^XM-A>#l&HPHesDYZ`EadK zweQztdo~*R!0#K6R@YvS4_WcFo=`u`|3DC9>C5DXU;1b7}PA zNb^kpr??$<6`$a--HHvytwQ(2PDuPXt%x$$Q}J}T7+)AEoHB+-C-nZ9OvRT>l3{V= zQ|B;a-Nn|&Ek>IpPif!E;JIxK7{RUCyczc5$=lbGV#uh$Hp3%GgN5?vE1g0`S)vpS zg5HF{CY@bUTI;}+x!xd7rqIP$S=px>07pQ$zXnaaAoN1QsSKmPDNYKwuHWB)GRqnN zWAwWYxP7h&83R2+ch?F9Ud722*!*Ntzy=>2t_P~vHkeJdcHAmD6s=3rD(9A9x`_CO&{3(n_?=*Zsrzxd=9WCh!uU!ARq=Z_wO<- zW&4qB_6wpA<)OF!dVRT(U*P4`+!#oP83osA!Vce3*|jyKDOIAJM5q4w+S$ouVvmK7 zg*Tq%^>$)=uzN!lv5uAuF(UyggT!XQJarBsj9H5>LByv$^dlhR$HUoABXPN@_o(fB zT|GG6$jO>Xe|7h=uPMW{IY^%8rvW_OLaw*Zy?^lwlx+?*ad;%MJ^=LOgJs-O(aPuE zfkiRSn{c$=kzqVa_UY-H7B}{#>*Yj381`*ArcZiSq1w=-F0aq{`3k4qLEyw`_7;I29HM8dZ?E8;U*%ARH+~L|pB-cK!NDF) zn5B15o(K+s<31BP*(Evb@2f|d8RW6?&%V5Nczz`F%J1jGeNT%Dzf#k*IrL<~Y)-UC z;2t*j?1-U+$z^O=0KsRk?jAVSyD~81HfWVtN^SiVV-rQ$qL)7FMNKd01D<~M_ z3Y*Q9-mYp7e*8x?R>YwU##~%~#l2iQ)G0Uy}8)Szeh z5ugV0dDVyYwJXH8?8^YT9XCS{1p z1cNA+bZ|C(x0fBS51l}!*2ix7i?9tiS19>Y&ljK9$_(lwl^W64p;m%b^qO6hKybOi~KZ zKaFcXdLd+S(5vylC)s_k%>o0-vSHy1hMRv6(oFk^@X|Tnu%yDj7;@5aZF@L#z;x)n zbLWKH;Fqq@QU3Bq+1qJ}GC zH|1bu_~#}0K~PEvt$Br0!0zUOJtbcq<#bsAMqi2L=%#>GX>XWTRuyCD?|up67Lc?+ z0Sjp6t@G-Za(O1w80tL7?~Ioj!nlr`1HsLcLrHxPv>4U z?QO%BybyW#t76JM6b1WWlrs@5ls5r#Kr#9*H0_Wi889@o-E)K#eP@{cmO@8`%a0rW zQacDKF0X7Ygv&5I3_rNrf%sSmw|HfB6v@q)eRQ9%S;C`u|8tQAO|7$I1Lt9GQS=10 zU0`sieH$w`E3UJ$|1-o!eoeFCdX9URjFON=M#H%Wy|UZGdWwmCA%GzA`~KorJEl6x zDibT*+sqTT=CQCKuGd8RsBPQDeV9+22~Eb<{Vd9Di0YV4flR+R z@nqrh|33JFG6>j=#~r8>d5aU1p_y4*r^;|sTDzY4g7HFcY4(@eUG50j{&mS$8<>br zehnCBtW3ax{pwj(Y^rKs4wvGRdsl2gsL<3YT;Wv*KkJjj?^2MR!*!~VgSNcz6eb_| zj6Bzv-nz(w=(}I2UWnJpk7z|hbA>VU7A8fFu7K^67WSv1#j-} z6j7mgPT<^vmnir|oa}ud)S#C-MH%5T9C0?>M^ggr0taJk{lXD{{t`oJEr^;0;=%x) zDPxR2k$O(d1%dPfv!$%MST7SHPB%Xgjp5b&TY@D#AdMX^A*fvXr#T*m8nB;lBCm&5 z;ufI}1Z@|`t>zqK;-;fNKA5BEkDEPZ{bXEKqZ>03p^Vq{E(`TJNJBo(p(6?Eu@taK zofr8iT0@Aouk@{03lWt7zkt>|i7O!--nnOm!OaHq5kGkV{N|c*EQG*efeNv?mj4N? z24r1SYWFr=0buiPTv`xmdbX8)HrQE{T|z2({x|ibZ}4KDBJk@QqJ+pz7dAQAa$Wo59<&rC@TOkX!#B%ll191m3l+N*qf$=?mw1b?O`K88(`tWWwV@xbQj@6XdZVO$b8x-=`Q8nD@O()LZXl1w4W*yKXi!U_fhE z(RQd6{5S6#b$;Ia<-48LQ>81k9pN}I??q>=R02D(DgS>f>h#wxRthb8aKG&<_!8BR zT&s~lKvQbbtzJsKuvDx0G(es1IEcOJAaBj zw4znlM^3U#Cj2K9bllZe6aL0=tH`O95n8MR?Za^@Ky*JE@u!;`0igcUZOo=)%ZIQ~ zGL>ldp?QPm_@2xPu47{MS|0%-^**?Y8+cuqi~sr?gw+17tU83(DEKJQMe}HT#{h5# z=xzfboh0R}393fbeDiy{*4~?|fL0exTna*}KJ-PjNg)b>rdP?`Z~2%R9tAz*JBj`T zdD@{wD!JpGGFD@xEEQb5#ApG~ zkzh+N0 z_@DY2H(!WTZ5nTVpy-M!7o~5i=I%`%E9(zP0W7731;B@%lL^E37RjlGfm4z3I^n~- zTFMtXPMpKBu}cYFyn`F!xdpxYa9abchQ}6VzzTmuRD-w=)phh`Cq@Vu>FEOG%Nn48 z&MLkm(Eqhd;0FM;lmxg)<_a{(Y>7babVUgyuV&PlcZ5>jp9PAws2_{m7bYY*0b0DP zELz~b{4=~VSavT63+LOEljpF^w!R5n8tXCRFx*WrYiYW~EyRw}&lma=)vhnT%n1hh1H&gvWV0)URjj4$Q}K5Uz8U-tvS|9+o5eo^lpAsz zEF*9ORhTjD9iKoaj+m)-EDP+EiR6kvd}EfyUQPRzka1rLp5nb~BS;CEPSRMH0keNQ z9C#pml@C;)XA&_-x8HF6rEg$>DCob@=+AsM=w;oTb=d*8XNv2D`eeHa zoJ?qp16A05&x9D-cl4~)O;m+m6w^xL!p4Q+DQq}jqZ)aQS-NupTPV;>pt@i5vk}{n zaj|vQT$X=_4fVf^!yTCK0Ft<*`KB$v)IEf0nf?^{#oLc>X$--NF(NG z*$~K2l^d)pp?M@%yT_f@mysj_GB#Pz$1%Xc%W=91e`sb7+Em}!x;n#$`MT9|b1no3 zG?`q$_D;xOw5SCRCYVAPaxgq`gtxl&~_V5Gq)<#(d8Hz zH5!KTpj|_i!odx>Fiy~w^{bGw51A3~wg`Xw08z>_sSylGh7^=_nOjT1Q26)Kj?3cF zz){4CVNdmV8^1yDw50*~i zY}v^Yw&jAkoc8YuQg_ozqKXCy-7fSh-|=By|el zdR@p5jp2Rj@WQ*BYCunCEe>zFm0Hb+33ofD8a)za#M%L;)x=NmfIxkawgt;x_|gUk zk#F4E53Nx^BQbsbW{Vx~v85*L7Zb!0Pp^CgMCK(Iz3^XF5Gx8IMxm895-Od;?keqf zHc6Ggxb-Gc;gA!hA8vlTrc;O-(lF2tIXRrYoPbYqzJsRx7y)+aEzqeRrenSEd6QGx z?WuH@Fatk(rMyYf8OAsOF~t+B(t|UKt-RQ|HwY3m^}ZN@sqKHr%Ge3_=5xaNv_|-4 z0d!mtvI5W2^DXTIXhF8*-U_@HlN{={h2^n3l}%fniw-GS4MunLCHYe~*rY=laol7@ z{49N?M`W(n>sIw2!-ihq-=k~}qc8(G94VtH2HvV9xCYJ^FM;dCGqRU9Q_|MY#ZXGG znVrGTc0C=UEQF+mb|ib~`S@tP&m?g0V%8u~fUe2!{5?#Ubd0nRxZyZ!kDUO(Vx&D$ z{)qv($WGK#!>~kD=TvPbt@5#h?x^u#mmL2(>fUW0>fZLlE4+aB7=)bK$d z%6`xwyn2WP(4hLrOHuFog9efE@IQSHceh+{TV!rOukHc?@$lY`@3*veVT*f=fp7n| zSS!?9!$sbM0Kl+E-g1sF6`5p4R3rfZ-qnikEfK0meQ-k;@a1`xI${iR#eVa0-AE2s zu{O~1I0|)x$#+v%tSI$|Nw5Xb{tPukV|d#$UbL&xB^;vAXeU}x_uO-nF8@?^!HR7GDc9&gwY|!0(ydIas%y<_9rxiP&M<)g zl2J5nU;B1cUo8VgTM{Ks8b+u~hi++A5wFl`iBnL0Epo7j#Xki1pdtDOq~>?Vc;5(x zIq@>2i#?y(Z=B7M$giqe+(Jn!Mud~c!R>gR9Ybii52JVKNna3*+6R4HjMpVahq(5k zns9<~jq4wxNR!p`;71hj-8uDY5{#@EfyeU zpz^S3kZ!Ve7Wyieag^WYE3TxeU?Qal6+YyzXi%aU+;;g>N-UraFG}*?3A!;i^DggD z4OO2&GvpsN2GG$Qp_W)JlzYp1sz{cHOI^5jq`IlZD;B@Qb3DQaJ+|f>F7+e0seKf! z8sMRl(^Pw}uZoNcLD^SvtYY&$3Ow8V1JbYm9#t2CgpZ=;)@r5tH-nFsgbfpI$e)T_EPO;YRt4o6J6D7OLNjq$5 ziG`UDHkz5<1Id*q%0$*cqun@-tWWs^UuY#>&k_8RzAe zZ~gC_)7i`&`JZ<3Z$bqr&a2rDqeAg;1q`nzp5Z>dg5>)q8wnW-m4|;{Q+j4@lZJf! z&-2_8sxq*eiv_VY;^oJt{niwKT9&1y8d{wPLiT#Y;4;fMy{u$3kHisL%`+ix-X=V< zwyIXjfB|Wc{#nH^R*m>@u5@X6RQ)z-Grv4S6497_@1W@oV3ZVaQtv;y;iR%a$eO=k zL#&?%4qKn15ASef+Uwgz73*`Kn6?#+-+EknNV~PK4lKeEGM}Tnm$sZ}*jVog*bt~i z@@jk}y1yowZjRBu0#bCva&lBFSaoL=*FbY{RDp8d9U$=f2k^kHBVP4Vf?!0TtvOMU z7<$+gMqm;jS;R0}$g!(dFbB74Tvsj~;Fu4xRp1bBwJ0K(VWAd#w?e&Gh^O=;3XYJx zUCCDgry)Q0z|hS43Va=f%a(lLDXK7j>i=(#S8b5>W|oA&5dyz;*C*!U5$Ssbf=>FN z-(aY>*`+FB{c87Fi?F0nX0ZBmf#rmp58-zk;e+OKY8i~8z3YV8Ka_y5x?&QdLvBH` z??pX9(z6Rql&ko$DsV=}2FL^PjTW$MKH#tOgWe1*{MK6%$kUiZo^cAkA6ZdP{k zO}lf~9tsztHMtnG4)7fv$qKq;su`^P*arq>$+g)>O%pP@=x_}^AjWv-W4Jq$(MD?E zNY8pijJLl^3bgP{qZRhnJh!LRIb$gxi=yrpmJ;bj7@{U-`8&GJT?UqLSdi_)L)^r;w@1=muC4YB{+b9dS zr%7?)PGiGBY(pOT15jw$VFH)T8-?M0D4Sk*BKP6coNW_dyfZ8xHI8)08x6!lX7C^v zvJs7R0?PWRQNAP4Dt3;}&IQB=m@40?@6M}cITpfcp0dFL2mLC}4)BnwMhcDtY?3^J1WF(J|sOa?(^qI6ifw>qR3 zGAUUSjr>Hs4l&Lzqb68yG!^&aqW0jvdDNUv@S~sh=~wUfysbu)%2oIbqgw#E!>6oh z!7>X&o~JQyFe%SCb6|(lcFbgn>kTU9)~c%|8$Akg8(;w};U>mIdQ+7&3#-SrCaLL^ zA~5V@J3d}nK$zul%h6~o-Ko2_2pDf>S~6OWIy}7kKCUZ!Mq84Fl4hr#QWkB1*Ok=e zRDK@CdIhF(S;;t!@vIsxc?cYe0w1DJpz;u0Eqwcv(9IQz>ENsro4=w@#NG^-R-(5G zARN!>(fXd%-k<7qv}_8p_~#s>$*k_WVTvkw(SMf77oQTn08M0Xqbv6#-1RDK7F0~w z0xo;dWQCW^5Tzo~9w!N#q5YzW?#{ID2xV`+Z#cGh1iVY`{QIv^;TP>O^a;OJ9>;k$ z9OeVzkcXSRGg|uiy1;dE8^7?6v-`PXg`V=O!^<>eKHa> zu!8G8vCY`!!;VB z@qSWD`s`Lt$%#ZB4+~AJaquh&PA~0}o zv4uGEMX4NY)oNyOD(YH76<2K@u@Z$_Ves>ICVLvmOABLdZdc5SN(+-*IZI33+46A1 zy_nk&H=F)|f-1`Td4~JSsi5XEVm4N*|4aZTI$sD7Q^cUf9qE8F`y2j5=ikl@SDOUG z>U9Y)a~X8<|ReJlIK6u%tEOP=5)-FnTq3Dv(~}kBkg;w>W3z7)oh^ zmM;8WdaclLXUB872idvb?!4Y_O@P2_ZEbnrModefl`WW+KvFxOoqqqx;Cio0HYo|n zG9smF;A*>|60Oc! z5zc;63(2<;--L?lc@ev4vNm8mLu)&D>^cXqcGl1Zq-P@XIDo-e$$baMr4aP{2#8YD zBV$cDd4`KzKITpLnHsqeSNHnI53CnrB#OyaWW`G7WOp>$y?=^$J!p7^g1w`Y$MAcG z9g%6*NYB)WOSrcD4Mb33-mFHAfXFqJ+?;*78{xP+Q{4yeZ(QDjlG8yyW7VWtJawuW zaLKSMqc^lgxg&qQR_Wo0VXdG~%5qhlY6c8P)% z_PF)@0gIaU(CvA~KggcQGmBm-q{;{9B(gxdQbGN6p;3+yyVm8V`G;vj)-WzQv)fnR zJDax(9v^>ueY2o&U#v?Pc%pan_R> zIL*#m7jz>GLZWsG|GD8SYdGP0TpH_Z`sL6w=wnE@T^h~Pe&uE>^DQ$*ec9L4c{v{n zU|!d)Q|kxr6eiG#iO40w*!AVrtjxe_t=MB#=gFL46)om6>AoDp>G4&2IW4G(dH>tZ zJK3T>rH6ntr=!9q>n%{+QgV_|Az3cG2e%_P1NsaCb_LCOScviDNslJ#7;K$8@JcGiB8No(Dzv!j6Ju|gELSyd3y;18}iCbrn zbG#i8pH--zZLE86h}-#A&`%V$4f$t^c03}Qwd$qaRAPpGnA4BO#|*y_x@#ur0?gE^ zO=D{n>ucgA_OGZyfi-LM4f-t#`y!*HR4_Zc|vpIu+s+xTQm)^k=fO|d@81Y@)oQKI=jd8%dgY$%23 z^aTT9BF_wjm&GW(X`p!^5{l)vfT|lkM@dB}1CmPbqD1QyJfu3hs4l?|vePbwyL&F+ zuzL}NEF%{$A|i`R{4SS@rWji=+#c`_u3fv&^srVH=-n|KY|3(mYi7j04=v-aStXcq z8R^73=q7?V`r7Y2SsZBt0i1n?tje*9*fqW48&vTfqtPnHuy;@=-q4oOq3{Ujn0ze~ zSML9s^m_8}3a*P?M8W+mmnSy;1p9$|rqve|HOlH^vT+MV)^B-s+H4UblEuA;flYf= z-vM?b{|<@#WW#d@W#U1=a3z3)^B|0@baZ{9=3IiAlG31A{|1;vTa2^#mui8D;I4Z- zb~LA(ao{lkULIin(m=@kj*5y5UN^7|72K9g&iUHxLnb*B?L6~!e8{#0qH4iN;8|0? zK9z_*q@_OUAo=?p!Ap*E6xp0bgT6(f5aOvJ4YGv*KhGm3mnVt^P|t!d3YRlC@?UbL z0e#om2@k(-l=T(`Yf|KGS<14C@KsT=!fHkrp&=eVU)wUV{*Zbs-w?SMIK zCGcBcS6gZ?EjD*|%@@U2bC0w|C)^4-Bm!~y54~MX-wn(*mA=?!1b!o>xewSMR*ni$ zkbm5m%li{CUU<)?Nt9qq4_8Yyyl9$P_Yn?=(4kHVN{}yt0ML1N&D=X$LzP~K7^IoK zq7`F7I~vVSMa#iG5(YI`s({_Wke7cvGyGv?*M{<9eB2L55*j z*?AU#1yh3>UkPDvPZ&mPVat2 zJ6x<5TPBAX@CQ;u(*9=lAFM0Ahd{p4|7;y;Pl{4aPm8kD?-g$|fTCvtDj_LpNu34& zY}#dba3NP@@)RoGdV{m`|73+}Ki&wMwZFmiMZwYJ!#={63|Pc)qD7)IPxN~dfBbk+ zsZHaH-VGK)Oz$(P_G0za_Vd0{j%7sai%AFtvjYII+75hDn7s7Dak!&bfR)NmL`U2; zc`Md))aXp^NcB;?kre}dA#|aVKy?(k>1*R${^b64iXbI%UP9)`DUI$Jl#6Hrd8>;| z>IZ#d`+uU5Ky?^(@IxJX!fLOYjv#md4=0m!I(n1mht?HrvVcUfZ08Gzj66b#$o_mm z0saWoCJTi#k9ks6)YUGgu2yEc%4-@c{K%TYpD~?u`N`*nYOi264Tv=-w{Plx=UZ~C zdWlJl+zu=jkik;qzgV&aw}_tIu+yN&tg@=F1zwy8xLR#A?m(M#n{Oa1UkpuxqQK>X z`7k?h>Q$LTb;sV6q|F|x9Ay1^DjR%Pa}p-@0wT$dP{8IGoHRFni9{btYoa|nJ^R)) zwUxCda~jrdzIV`q2@Ms!Y4!C}&b3qX=*|n5hw|uHKJA{{rXdP&t}?diQGVdQ;d%> ze%H!|x`K*qbNhSoCO;D~BQbVATT=qB!G{2j=s|+dS|S{0ypX9MFyAg)Wui2W0p|<4 zOc`!}@0g5+NiHN&b=sIncvVB0&Qc)c0?6B<(U_-FHH-aVSSNIQNsKMKxHN0G!j}+c z41B7hD@Wc#4JT=d_BbFSaZj|gmNpg7iAz9Zv?gHgpUg-p5>Q& zkMR$R;h}0Ifda^YyUEY)FQ(xO7m?0oMa;9X=t;gyyl-xH?xWoE6$fj!#q1=3?c9%o z^r*&B_N8Xzdnx^)NowlA=+(e21)r2UpJ%$d6tPPwNdmBVPmrE9;VL@vIW!Q8D;g=4kcSDCn8vE$$4+GO{cvxh(U-K_#S0? zXRzvj&Iuo|pS_V@`3ly}8T%B$TXJp5pFCI6j*={Ka%b9J+tg2nLtqpf(GXxw!ZBJ{=F zL$}UfT=*iJ$5u7Zz@G=-6(>T&ER%|}(f4C|tD$>Qqarb_v$8W*T@v2jgb&-IM>7CY zJ|+)eklh}@ur3rX6`S`qeJ&(Fzo;OgL0B@dtG9{84Qx)5Z`hQghKvfVS zyfsg*H8Q~r7*kCrlO&%%nZ#Qhk}x2Wk&NRFK(V@&dVthLO43!R!#*6CCpX?F;bC#w zG_}E^JX5L3q6Ts;!0D>jF@JR%HgZ%=f7o|RzYF#!dSry={*TRx<_$6(aLE5aw)gN$ zoJrTu+|!-l+)uzsNDz)N41+)08qUZA^aj9Vz#Pb+srxfnJG~kdGbdNWUIUC0v=svx zBukZRsJe+?iZRCtB+n&8y8mSms|f4BGoeNwC3GL zig`^Z_Q)U{iJ5$Iby|>SHA4iCocTU8_@RcX@?n1x9y2QqUET>nY|G<-!)?cHMAj+I zMU)EKWS;_j^$dlpX#&ztZVh)$6M>VaDdS#W+|2S@tUx%8n zQ?E!=%Bz;7kdXkF>XW4$S5+7fE{MK^aOVSFeUYT>ZHqi>wZ}UcA+Sj&8Q=KfHWW(H zwczDX6Y_}yc5c#XbR-V8MNxe*U4s*tQ6 zCz=z`RVcEU4E1lz94NyF7a@?@V@#~S^MWwfgwkq#`Y()Lnw9(ubMzPX5zhjq#0Y4fB2zg<-oX8Qq~S zevm-XRlsr7WRA<>JWkU?p9p)!naW1LMG{;oCn&O*r-bcM@r5yvGK?P2i_(Cgb*`bGZeQlcH`G&H z3wF0-I)a#eNCfC;{I`f+T*RfMe*#)NiGiFcPWbda+5s^B^4*=O3#lSJ{N_=>&3#b&#E0Hlb z^j;4Rk32UJFKh%zc0^`-Vz5Z$%pjH1avXe!RwKZM>F>)NK1aNR88C=^I= zvh0;lu*GBk$;=h?Hiv+Q8wL%d{1bGa?jvgw)C|wOF_03ngVk!~p;R6>YsplD$l&Z*aM1Z)a!Zq8^@j>6h+(qB zSS91*2rnkoc#+cdkJZEz_sY*lUh>)~+oT_72h$g>E_zFz1N~B44ga zF6Zi}IpcNvd(D>9DpL+gq#HOIcbZEuAQ z>Kszi{9^Dv$$!9cE~*${Ajy`_#f{@mu=P<4(MO{v5TWUKvHyeskilD#tQW)St0pF`&<=pudHOKI$#n z{>|_b#wIl++`zeQ%cqIU(f(?=z$WhGHQ-e;LUf3X0Zs5$T96flnyf=lXUPdx3o$S! zm`XY8gKM=j{s zRl8Rz$3n2*R#dm_%ju6^5gD_CDNrVkEo8_J7;K~@k>{F5zdz569!a~ zkxK?^!!^;gyeXD~GAz6v+Rxys56s7l(K*P%TIZ^Sj|W7-!&tqa`N1m{th+zH$mp_y ziD4d_O&x1kS~}Ut4q`Zu8Ay~Nw(g%pLgn9RU%slI(5gc&rii#xI|y7oR)6yGHV{v> zN5mQ!mS9>t6(Bd3QXb^>@lK_Yg~F3gMK-2rnN0OK+H(1#k|f;IN%0MFGWLMKTCr}0 zvT2zNp#t4zx&Pc*!fM(Q&@Y*B@vrFU1se9}GcoK^_D0MKyl*`sup2*x9d$}G_`{MS zgJEb)(<-lU9PZ25p{$-JwPe1L&{>(QeGq4L?*z?NgW9|aoIFvOEs0YPcn|1V{i=gI zkgc5+^i{dV#JHzIJ(s}{{Ev?wJjXk1SS*W8QjIhkeIWwW*O5o}xD#N!ivJr+r?;&G z%!E?SNq#AkUh}ubuQL6?azkTPa+^Hp_AhKNGf5;i_2KSxV9GXs<1=C+-f>%p>;8=F zIN^8J*V9gd#N?V_;vI$H&xj+l#Cwp`w9JeyQ_R~HM6=9?xqsr}Lg`1oLnDoycTV#);IOw&q06D}9wV0I_L27*wQXaXm z!P-`zk;INC_^x+9FWa=kYr%~Z0R+%QiF?%ssWmPsp_3*_UefW66 zHRn*vFBT83j~0yELX2yqvXG9FWJ1@tL8;&WQ*2r*6Zd5K0`$WvHPcGHNEe%^A8G=z zJ41FYAg+xl8UmNf3>R-#K?-jMnAa>+6 zZBXA*&AO0W@@LtX{0@WL(>aW)oOLXgq38@x+yqK@j0tAN@Khh4dmXT(CJ{9C8^#0n zrIny4@2mTK7p~rAhP#GGQnPk~QbOhgDQ#C+rUpi}*m(dpazK7ngAVTXQS!~*R=>q2PbNL4s~pNG%V^H zaCmf5rlDI00=~@V)#9mw@q5Yqf?1qPFbg`T(#wV~GZ6r-VBMO;4?Ta*)r8tD5Za5D zuwI$>M4Znr6LURZDzs81vVQ2`A30NA$u7sWsVGFX8F0Ot6&&&=TbSd~J!y|;*El;^ z6frS<6?VHYNENaU@nLCAQDrk9nD$hU2s#k)UsQpNb`B2Wd3Uq!Qr!_qod*w4KjtjU!AFhy;xxB;x zc@HO)o}_`q z*e)!&ikB8Y$W?@xt3pg6x46d&2_TDp`4Gcrsp83^ScBs=6_H}g?-c@@jzhB`53uo2c zO6fb-J0Ey;LN5vq9OcWVmf;IqT3#n8#+~$D1dpT}_d4bCIu)pYMgf!Zp0Ur(*z`G#4Lu`(`lWoryu>-UDDIxa(&Gwzho7O}|HfraV>p;ziE1&`e zwX3nzyY7aZ9_g}w6{JKNP)PWvb%UOAmf|)S$HGyCo zS**nb)A#;E4!JGYs4WEU+xrLoj`?0$R_g4J+Z*5=>Zyaapy&njdyC zP{-lpl@dzAE9lOdzt1Z>^9;Dcoe3e3e20bN(kj#Hd=5O^B&3xd5OGzwxn~G;6AgjE zj%gpn2{tQxhA^O=z_s@P!?);$nrn<2bB3P1c+;5Z`V(S4IBE16h^o6{G}R;BtQH{J zUSn1DWr8}0$#Rn3=L4LVv8L{t6_N_LPMFUZWY|6M4!=W9>jXKk^^h>^HkjHPU!aR} z#&UWh6k0}zAe(21=~DL#w}W(QM3z;=7p+Ti z+dc6#bm`tO&~MLs-BP`S`e9V)f1w~v@zO3Jy+rP#U5msjYbUBcp7Iw?y^KrVd@Pgw zPod66E`RAYF(2NNu(!gJZ9T*J|9g7cDzV9KsG6>r``H1zgl0~aIEJ^EcL#2mo&j+V z@4K-ghW*+y+W+#ZT>t(YCJqsGiP02v8mi1pfcwisxz|m1aC*Wfxz!I!b+qtc6chT< z{YHCEBefF)#Ekwd8}G3j;3MVGjk=(_Xy~uDwpBe7CfA~p;q%hIuQ|TG)Z(HS5y4dC zuQ{lQ<5}CB`f)3v*Z0*SmsUe8Q)@NF6zG;#ja4+S)@FwGX2t^8h&GXf?Gd9Yf zDCps%lt}ktk#?YTwgt+UAkd8o!x!nb-x^Yi%SP1pddjs#S zHjK$49HXK49!KeQ=g6dt9=PSNBTA3I}NS;aHS`tHfA{8e-FMu42s_dY@Nt zm0(M32Hx5tqI(X@dB5{}lm&HKA`|KQWLwi{%t!7TLK8BDaY^lyy5jG~<=!&J6Q)Gq zVk#i(`+v*3N`*gxsYcC-lMypm~cW@P>`K9 zhSv6jZ^Q*6uiHn9x~>FMr9>)S<|eSV3t1T!%fxaPpMem`V!u*1kV&QXpETOq9J#Bx z55d{2AJAc{9(V&)8Bt@;Ad3w9Qd1lRZu@e<1rtXu1T|;JorM<1hr?R zK}J?uWfdIbr-A#QJnu>rd8Td2XVA{vmQEY?$+PxCw8U*uPK*YATL3^aXX%|xjU z9xJT{`~;?y=T+D(Ujx1nyUUDnhy2ZA%XLB|BGKm>yS6nhG3VU+O|v_&fAQq(0uUF& zg7>?u{)+RyK{$32(5iULDl^Pj+pg^!w<8+S$IlgL8O$aFIGb!BuV(zBBxg4i^C7^j zs(G=&!9$4KUsf%onRJoywdU>H)rB13!Y_{%1Td6jTQ+Aq+4rk|oZY#SMntwL3%_O$ zb}$vm!{PcRFvk+S{h*RzE2S~k5Wvw7i9ub6>52=}bzaWY|7YaH?*EtY^EJg%-x}5# z?}ra;vo!w2Tk|p%6zRL)p~l)o+Yg%fc=<#Hf2-iUY+yz%hGh2&1nVZ1Tq%DQ5}pp7 zW;Bq0S8Pm?gq~?6tC|MEjpH5Ddv%~vYA_E(1JSI4*oM?D7wTwJn#`e(TQ1QuRo7s= zb+ILb@k;dZLM_|fiIK_OC!c=#t-zfO5w2FlT7ZqnC-=9cS_y91D-Z2UYp+5(UAt^Q z6i)~AP@tjF2PmC?tu5gv*3f^0v!oD zCnh<0$a));jPYOvxDtG4R7(}d-y65Y{UZD7{cVHWl^hz;*KpAI`^?w=#T!>+l#^IA z@=Wd~ClZHH^CFc-qqDm;NdHrLYWhJ5U_5EENn%$%ztuUN-~SU?umWTP%pT&1J%uj< z4Ti=lxLOi#Tw_y(ECdvhf|HpyP{)+)#wNqG3|rm^cX|2zDHjl#2lubs+UILd|hS0^#&z9vMLO z$1yURg-K{p@TGNuk)BqV{EdYOJ@br^P%w8nSL>~9riK79k%4;|iC{ZGn)rH=ykYqq~v75!t#UAz`B69PXuXS(B z7ctjgu4|R@-5WM9qKmKZ%Xz)S0|$a>BqAjH*;F-RcO?{aqc)FI1NK(|eVeHHXpGSC z07pQ$zkes9WxXTqn;So*b(ZO&J|m9w2;fFaNxWA}S8L7vsahdTPWZ@vC>6)}%M7bi zjm^3~`~0LfNf&&-=#~wGe};Xkk;pgLNnQx=9kLUbxQ2t-7+u!mVStCk3Se<3Qb-^_ z3`|Chu%YmC?C%G2rAjf|?icOv+*9|(IDQ#5P%xR*myN~LZVylSgoZZhOUbs!bGdZ;yfGctgjn`Msba~rF;K;V)1<~{hAvSr0MC${ zikWfv@wslRC3KTnE%LF;t#38AvUIZeDi{1;spaTc^Lu{&}_tmvV-vk6mDxR2Mg&&HHWf5`#<@yZnbC-xOkV zf}Cp-xHOui!5z9h4m>Wl6YCaSr~D5_vW?y{>8b4Uu8}vBTOK_i+v?jTA}Ox0ZJCVE z?KFwC0BLZa*MJO~J;Z?bXpu3C$HS)qDbhs**R zyRIH8c%!@~1^sYE#)v6Oy(3DNS55#joIi~|5b<$l)Xv=KJ7=`hQ(Q>8(T0^BkP=a5 zJUvU`YMt>4J(VhlQG84pCs6o@yr>~KCqyy_#=c(-l&37;qHlg4h}j3X3fa>DCE@%3 zND3?lA86CDZK^C&P>`cHbfHT9CHDLT`Bgy(MDl97|D&xI=L_%lSLb5}O(ozE6KYw^ z>4JweNaMTQw}8MR2$k!RkWPO`qO9r^|L3yJR9PR+x6>8XPq16EBpM=1f$(e>QPGFX z%@#z-F~r=XbLdnT@TJQhGjJc9{DWhh^vA&Zf)q}>#Us|+`8LLjq@~m@O|tYn=tzUc zZ?CKX>CtsOxXV--`dmR@;)!*Hp9k0Y3SyS^TL&4yV$1v28(0U|@|tyJnG*TGcaoK` zR<>Yt7FlsKkppL=VEs=xTah}h_mx4mclNjX_v!pI>y85$TBj_X%I(;E?;B0Px?VpK z69WLyq!>0ouW&H%?Y7*`ne`KI#GE%GQS*JG#+DYi`2f41(~XMAo7eT$giMj2Z@J_N z>d5JgI?Sm#PB2FOCkK9vg=b*~6=|NL2B)QgNHxYw9sp_tckbsDwHePeYj)MX_<%)nAlpS zNM3j2B(!uSu#jzl-I0$guZK?>7_Nz*I$rtxCZQ?QC)O;TBc(EMV+Xs8uo%j?Q3jZn7Tc)#=KKZieE|qh z#ZU8=IaC4bl8`E1B14p!fd#ZY>-$IgQSP-f2)dYXj5DXbL6D>`58c2Y-Ns-aSui9W zz}u1;!i>tnNdKduBRd7H-l>YBRH`r0fy_0&*o|tS*p+7?d z2vW?p;O0z?lVJFJ?I4^-bDk`b-Sn@&>bV&$9!i;9N5pV=@uJhgai``Xz;`T+W*v}4S`CAu{ zHP0v)b#PKO_G@gu0TX=pvv+9m#^W?i=KGi8-VKuH)FJcH^x$s^I*IiNGgq{FD6WfN-=dl zsP^_JASxGcw&O{xKa&bJd~0ixj(Xarr5bQq^)O6S;QS_^Z|7^d^~D*zzA4DU*svUM41$%qc(I~ySy#JjaYOv6Db7Eh z6+f*;^RbI0Wf{{*s}x<1Sq^E1-0&0k?GmCMTSC!7Vd3%%9pP?nv&ypT(nNg8Qf0pA z2b8bq8kYmxOk;-jUyI3_fhgx`Ttf65*2}nFjfz(v(1Is;&$iaGw=)x(0TmljaTjHP zAby!O=tlM;hC-d_tkw;L^AVq@EDh9nnK8>5n4OcqZN06l<=Y_5kMm#(1f46CZ(}2-&l4g~72NLalYROmH zagFf8rm!NinBBpW)4a0-@KBeDllr0V{t@NyRDA$!(c_bWbBmKFLF5K>xK{vMRdgP zARZNGWcI@ZC@#Qi$HdQ@1Mmr_!^+7E#}#VwAL2nJ5f2kJGdV0aQAo<(s4@pyo=2wn ziQwr%Rx)#>OgM)+byN!en5qZe{f!f>R1z5@2yU4oT=1sEH|08CxrqwG4-85w9>V~f z(DIo=_+nVX;)KQV{G=j@8^0;b9td?93`aSVl>qN$FXJgkijnI1iAT;=&T5RcW`!<% zl)B)MAi0SHKDMCO{F(jCCrm`Ogqd7Tbag-tjh|Dm+NarykYp3+dwcuD9f?wFeQ6F^ z!X03+aTvyr*&6$IPeNWEKZ*1#jVOpF*Wu#+k1>n9N7|R0CDkM@O zJ1MBbXH0Ck{CUp{Zq89#GJ>-U??e7IXdh+aA|jB26ve0_LDXW}hltrLMgN&(53r@D z;2mqj*5kl`uK05%@n6(U#*lWrHVY(Hmx!;i14`_q?Td0!UR&tn{OL`+8kH&dwExnB zL=r%%6OqeQK+djVXj0;7zjphR$Mi^2DAT!SVWV9nRdC$xuUZ%C+nxDvC+P@?VKIip zy-aSsC*@0mCoT_snm)%PMaT(?88C}Q>-Jlt@JUT%9pbprT9@C?7 zfLzSzxYS?)R9jHsu&qcjGU-;1im|~y6>PUqj-EZy9sz1^pszH57V@+&E7pXkyApXl zbFsYk#ka3T>!;|rL_6Q}vfCL4jlzb&UhCd@Pl}^vlekB3O9dsVLK^^|HIScr>1o$d3Y> zR&zEgS|n}O6?JN^53U=O!T@i>RN;mDeP?J=b-P+rt$Jl;c@lzUX)Mp-eE3PHG)TN8LejWprosn6xGIMP&K{i7xPudd4 zV#CZs7uT1r(|DqyS=7n-Vf=i2n`g(A?&%Lj?l*Tvu^Q^Ebb}iuBm1SI(MlmUH|{ra z;nPyR4cl8>kxO6gGOcQkM?gpQOzpWn*Hy(h#;QKXIpjYvpKcV{%eL86_wPf=%eG{A z89%t#xX)GYHRERx{Ps~>{VEn@)613;r>MNr7xKR%kyq#<3xI*Psfjj3YSit|^leL* z=Q(RN;JJh294#37viKai*y9ejnkV=zOX{s~&s`jGkEyte+2ycgQ;=bilC4veF;9gd`z;T+>HxadDFL&n_n7)gncSM!qL$H+cT2= zm+k-4g>D8?4z6ma`9FO|x{9hmNaXcmT09NEOA3n5#`m%VW*0&r#jlCxYoxmSAUBHY zdtuGw#GROY@h52qgjBURbz0^}{6BS4yb=PApE9s}o|g%9y=rfzdpgf&_i$$uz~o}D zot{CtM1bsxxk}Jm!!K}Y;(p;gGaKl*O4GFIa}su#CFtHZa?t13oAEVcjOM(i-=v)a!2 z1euK#cbx$pp%b=sZ&3WWTE3n`vLG^dbFmm)SdR&w^gO|~J=f6DUs zvXr(7#`ZU+e%riw>5T%}qbF_tCbqe0SIwjSBpX{_AIBe1Zz;Y_6YP$k>EKF1ViG*H z98mgj6I=v4qVbX{pihbciZRuz)hEWiuPr0dGJ8_xRD%?yN7v^&V2eFS_=3Z^-gj44 zS^QAI&)~-lu0I@<$sRcn$lPdJAFKe@T44iEgzuS@owQp4L}g7y9j|wT8qDXybvK7h z!X~_CGE`S2QAnerw*4dw4lIrV)QZiaQ@vY11I@&x6lBJU*f3KHZT08x?KU?9sp&gr_Hpso_I8`blG4%BmwK7Ikpe|S?epRX-=rqar@tP( zjH|3ec1PFzPJor`hN;ekxV)-!db+)J8S+EU+Tn5a@gJC#zYN$n4D30`3Ox=a}U|WzjSSZ6WAj*Q8 z2eug*M10=y))UjxtwDY-TXFXatta>VzNW#ij6+YS zeKw(dV8V?_IDtxv(GJPst{77epE$)R$SC4@sgB*f_ZK*zG;^V6Wxcs?0^ zoNW?T;=*n7#<5*2=%J7)F(MOOkk|!&gbyb#5Nb6J7ToWBpJNdl<-?hSNaE_HDsIfB zLpHn+Aj8_?rs*M7z5(zu?>0@>5MXpgAKTF0k--g0~u&_L{FZ#A25%)MH}&Z2NjoqKLprnHuOR(x^Aw*gXiwfi_} z9xp56V7(_l#>+YYO4o? zg5*8P1+&mSoIbV9pgktuo@pop1SlVWk@w=JxdP(jfAi|znIDu`Y|q+095jCp_Z*rZ zke>T=`%#*+0TU@n*3Nl)( z>kUn{(*mcnn=p+CT~6ZLUnh!viV1)P30evT&BD2XvCsE^ZKx+kA-_@&PE)Di4|4XV zDT z#=om*WxGG`jqD86>+1pzRn>AXpW{y~`4c0+r%wd)O%Lamiu*XW9-dv<&&=HD^A+6|PW2(f!e#$WFG3%8GJGy}zD~F!ej(6VqM?Dfi zI%@+LCNxXqxc`Fh=NBi7r}FmJOrm$UOU;2s7M3@sqd>33Jqbh`y)DWR_tcoBn65*O zf2}gF5_-oSK*}{q6HFj5xannU{vgE&vf4g|;xfw6Q2*T)b9xTkV>9$VL7vlR- zG=?_?3mds74NuZW)5iT(1P2mytumZ*(JXqOlC*+X1*7yv6LBNlznyqAhF})UNBjh{ zQ08^0!JvnLJV~x-l+$8~xdeC-{#PIB?LH;4U}j2Nasz=y+5|(=H`M9x#le{87hWqy zV_&6gPuI&U{-js1Yyn3!+fUpP=IjTq6+CU0OP9@0V(rH+tPfCiy_h89u-h>GbgwPX ziX<-ag>>9yXrK?kgCx|gSN2Qsqsw~8q8NKoX2_LGp$bf=>1^_%)dUB@kSd|MvjwI) zE9)uOba(2?X3CBcA-)d5Gdc_5F`PC)0W;AEw5t@_1tU3c;-#_}+s z+|}Id9p8|AQXk55m0IiD(uISOu2Xd~nXn;L{eg~DCt%*>uYGSL37^2Mmk{GzjWY)t@z)71a>1lDbN}p)u~CLq~cc z&Z+MdTCCAx6QpU)VG%vd9vSKuyuN(Q?`10?B>g%hM94xxbY3MUHIu#QoVRn~rm}F> z6ca(eQ`T{37*zZ@pH~?U#ZD#tz=@2aa{*c6a6rj4h)J(hJ_dpX`AB7jZveomvdb(d z&yyyqWD$v$be~Zu#XZqTQL|eGqX@C_b)SOR*AkLKeAEa zD+67{eeUW#_IQegaCyk$dy$)F92 z3x)&}eUNe=PNJsXy@qUtBm%$Z)pM$FpQ1qppE#ZWDSIZ{w5Ui6ho$4JT-isJq9sIl z2>l@rQdHddm*?m4QloN>@2V#7hqiX-4TsK514t;5m zdBtrnbb)@eCEg0!BI;~B_ul35REni;q4HgZ9E2K!W(X(G?-cDycb8_I}lBQ}_7#FA=8#~DudyoqV? zUzASA;%a90S}7VS4Ql>l3e=+sGPA0Gx)y9F%8kGY%xv-}iEXueth(j5@WeDPSgbW+ zs%vfaPfu(_zFPAA?e+eytJubj9o0NY1$bY?T+)pl^{tnzQYD9NccOnab*~6h#(j3^ ze!TH-)TTULi_-6do=D3d+oFTdmxTz{EwpkSVdhrWhvn1^9KLB zdXRu0(SggQ`zONg7#(X}3&H+8Glrys2-Tra*AFqaTQ^Q;*zwt6^E3PfJA#bn+haM01t z8(k3cdHIns@{{YE$*9|%-cl4Vuwe0sTBctq=i~)WbZlf%9GxrVUBc3413yv?O9R6c z^(NAPP+n{?^BFx`61H~|{API5aMVCsrYZ)=mVa4(QzcPqM#)47_vZN<1J!fu-T%nY zAoD8;{@rt4)f_WTu7g&F9a94lL|^v~)CS9|_4))zu=r!aWd;IDCVqp4r&8-FD1_tz z@qfD{57&Bl|6;DHgTDG7NK*caW+~vaF~{-X99=X9N4GEVRTWMo-3jc4oJ5?@giAK9 zm(M-`$y&l?6ZN3Z;+7iboMPexi1gc@O68na~AW&AdU+a8H*?;YwRFxv>hg&V?q-8IaAiMH%1AeVwLhYNsM--dY`%P zLHNF$+do(kVptNdG+nr(WvY&Nod=f52F8Ro5opAnto$x|&bW?$IdBPI)sT1pjRyo) zhgKB(?(Vsw!!N_o6rZN2Na^)l1kAF75Mfo=YC{CaZ^O+nmu(AQt-XvY(B=3;T7R-nG<9Y4SOvfm^M?tM1xuk3NRHZNgHG4_6a} z4QNyW0$D0)dxbD@&jBl}oLWbEVX*0{FfNgZN(>alPnpV#7iw{z^AAvAEq0NqtTcJu zvCoUltp?l2<+O(^KfhlRpOT3a`{{6GkBj$LKqflut+c74ljhi?2d^(#)Q!cWq4D$+ zY-;BmlVg;`vrZZ#TlF~J2Qsk1m^Z|$MOStm{KM(HQA$u!{QT5(qpcYKJt$ONfcHWS z?_fmZpGi1ta215P?$$b9IBnC}J`)V}`>OSJ-NkZ83IFj0L# zDD8FO5?r8_YHXTT!L3|vc9p{9dM;>xeeC=MW*v~ug-jvPTdwb%VZ_NlN64g@x;AlA z0EvFzGKT{9v^G2}o(ZU1;NKwQJO?0m2PakwIy)l$sLbD=sr*!$P^ze$e$ZpMXRtaY zjD~mFYG`aT0W{L&Ny)x{t9)p2<=IJ))Uh1gAA=w#4sA@tTDBKl8i2nbMcXF855^yL zD}`mA*}BKLW1LS4)fYr$Ik7DHkwo5pkD=^?5T9rlU>O2&D<#OIOm~0z*J=}0MIHdEmfA>ki5Ry-(Z0!^_tN56eLf;Zyr}&hIu{QHi>=&#Z zK`a7Kjk2l+`c1-~l}bnzU=m^5EDCv|=llQu=C)yTArm zips~eaS66DQKA?(UGPE}q$~1ejHhR@Nsn*dfOcnhQjH=bxK}hTd6l4ccgBZz#3@;Y zPJ4Zm#!Y>_X=<~h!NXp#w;-pa8LWhxLS@rAPemarlZLLa7qa=3EAyE~fO*6jx)>ef zYuQi9+p%g{uy#1iy6rIP59J@(j}c9F)1we$CP-zLb@Jfn4ltfBG-|I)_-$M`>x5JI zv3-Sk1dJ7XZS4CU0_+~XpVoXHXUg`0^|YY$#}2tRt}3Bt%T0xGNcjTz{W(KS#ve%- z<1E6@Q4>#D_!!i;ImIz{_gL%(Pscb_0?rTfO}j*ot8W8V^%ZJWCfOD{NOQ+7J8g8j zIZe9-s5gg6T(8<;I#qz{h)12ec+*Y0ulFkA4z$sjsf9vkKSzyRJEr$na$sTkQ05ku zauGwU40a(|bPib7a=$f}(a;6ifc>;z2#G3tx2Q!6K~fwG2KqA1j5G;TzG96YK3%+K zV1F(4GlbRplvF?cFJCz8F#IElBVXONoq5l~eO;HHp;&L0_lGY2w@@zqa_2>#yR+a( zT7I;3Y~~l6(BjySK`G+g5Z<3iKr&Oqf(N@ZBplONNOW{=s}xftiKcKkY|kD-m9Xc` z?S@0-tv-{!5lIk(=RusC%WgNChB%leoUgHyuF0HGC0y>2&!_@XsBVM_DhizeSJ9E+ z=9x|`%K~r3T%!l1z`s9h(0m-41~}tfz`!_4sI9FJgIcb#n)2yy>C;5_aWOGwxW10s z9tJs>MBDFWm_QDT$51ySC>&@U6L;)!F0DhStyL3vJ)O5{!Wi)&dXbBLXYPic+};I@l!z@rOTU5 z^TWKOlfD#i6I(CJIA1x~ZPe*D`sh3!-_lNQhr}6VA0mYf{mFNdd##> zjMRKH%&;uAE)|@gibE3I1ve}eoVMaeJB$uQtbNc6CQH>?P~Cxp^M1e4ydh}wUMulw zTK+1jhF?El*oql48NNbi_6`;pS0D;YN4 z9*-p<<~txD&pCacb0e+ahxDVzenEQOV<6pf3ATSso*{|Xap@g1D01=VUSP?33Lr$j zvxe)dn;!aV47F~5Q&pQiJ0j7Vfgb(wpUo*E`@=1Q{=tibT@{t~|A&2)UByKrmOyY;>Z0b^ zL{LsB2$hMIt)jyY@ZzneF!=KW-^%%}mj6VVU=Bua6JcO437_u8t}P76X9w{1b(WRu z7d@h3jt|4d&;T#@B&ws;JTsac=MJ8!T}X$RYO{`USB0mHp1Rc1u?hthU35gCq95-e z_^CpRfz=L>BZ7KpAVvxQvP4e>;^BMF6wj2eaw9-Oagd_1sb9eDJE}|WC-lOAd;Z^t z#Hc5f*$r{Z&wJFZp%=&NgM)YCS-w~c_gYPPD%{QWz6uVp(*FZvF0Ta3FMR#5IF--rHnq2W99FqJ^PN-Fe%s8- zS*_=;ar)l(I*uY3H92DTP|ja_E$1-sq?Asj@&oua5w=cO)pQ}V)`QCn(WdGYS)T4$ z5WN8D{!-0qcF8)G#j?#zB@Z3?NNx9*i4UQ?kF4P+qDPb3>@!?TJa4@0-R`4=)R^y{ z*Apt=4r3Bvk{a`OMGk<&L#RA{nO_vGn9lWOzJK`4UOZ)bR4$n_N8sg9EwLf-S>h(h zfKnQL|5lOimON>CvpR^w{^NRx3^VZFJ9?iK)|?+OS2W3Gw*t@Io+mxixY#=E``xaavnE5;!XPnLuh%U7--C)#`J5Zpf&qUw(4|$MiL_ z2c!RJ_Cf|nbEr_lyDnmO^8#Zs7W;cIq?gT1md?0iB|kxOtLDY#CwLX-Cx#wJ%B{O( z-ir#-6K+_aWLT6?5TFzhDyWdj22{VSk;$01pBLK|5O|Z@X8c8+)p4i!-5%d>-VX(~ z2Kt}}?E8@YM`l1*4@v7_S%8ef`nauDeihB67lC&H4`6e?4i|DU%l3@axgK&aWY6%7 zFDjtOA{1fWX#zeU&68r4fYL0>;~!)e)`f-p$ye`ItD5>W@Z)c+1T!d72Tr!gCpmwj z5C8z`(g;*+3_USuZ%TIL*B0H9V1_X>IK)xEl(6F_`c{0{11;OfmM3%(2@TVtK!4(!hAx_&i+Io&EINQ{RF zzZGI(R;`%b8+=#5bSE+OeLpi7AlApA@y-klJe)DGgYJANZ2WhA92?Fnrs)`^m}w zgHJzDj{38_Z(+tQGuTQ1f1b|^0hzp^w)WBUKX++J$Rd>2Z7ZwtN!aY4H-b&R>i4H` z>`b_excdOw9N%lx-b7C3XA#V$wPc(bU$W8W)U;d(OT*pMEID!|+kah0^uvksQ`LQ% zHtVPiDX-t*t##?`<9tqWyQbAHbICUE%-l?7QqE>_$!=Vel{#+g;O0lDrE1MFPoCG8 zq7$m~8Q-p7j04dGF@h%3KwhmZl-4B|RPHDPfM+RK)zB%xuxUNQTmx)N!?L`RLcYmQcUgo9rv7*5c@+qL7B3x=!H$~=XXOhz?=)VF{MPWG-rPp3O$J**wZI)#& z+B{cgEte4lzcwpVGD2eWv>NlvH@9{(zR>stIGvGLi7x4N#BwqnW)j#$k*xoEu_3bj8kX zK>5*ZBG%SHUD+8gK6cYsW@|QcO_5w%_HG*Awg@y54uj2e-UM4IiFFRm(MNa4XE(!G zkl)Wb!B!P2{1KpL=AE$2&v}>%=nVovUdV(ZSZ^f*82pAe@37BM8rYhO1ue(zdfPJp z;vjEL%bzI3ZVF}Q7&N;vPApA0^dFYcNBh@LBb2Y=O6@L=>%dP`7xGh#Kr^@{caz2jShj8WMu!4zb66HN%P)1(Kj(v$kLMAY& z7DRfhC$rZ2rlaFdt*HvLL8NN?cfk}Sz6BDP8(QC2lm^B$Jhc_X2dk}Hb{M8ZE|~{} zNKBq>88duSS;kFazxtzRY-1*z6!UKg*WuBHVu2fP+LCv`V+99)hFuTrukI_zwlwtU2$vd_nTU;nPFyMjf;WuVJ}=&wOEM zam8fQZDl4c>Gg9s4quH~c+sGqWCxflK($po=>L3ZHs$bSF=3j&v1>M(WM`KaC!+Bc zGuRD+*~&K#=AA$|E)OQb*zHP+Iw4A1XfKfakH;j`FugRW=f@`V1W-T5F?q5wiBzWyeJ* z5h^i@nhRER3G22*1o^I1c!JPMB+_=!-j5I>G1?X&X4Ou%IOezw5>swD8cC{F?Zlbr zXtYzU`VOo{=&b1q7H>xN;RAx}9SNXBtIVUHsi}Q-(6{d)B6{LB9H&k!_-OJomk(vt7I_K1Ma0UObDA_SI0_GrpImE=K5b4B1_m=pcm! z%K$fuX(hrm_nG$OuMaSg8O;Ww&OOeHW0x*Q#;5GMXS|Cs_fqb`HAB8YA*Lv1VTyA(HTY;4v5Z+l2%#U$TO zYCF1JaH;LyC6fM#dy2EbLM~IyHZXH|MC|%w3dB~uY8~AssS-&q5zf;Dw^oD-Pl=aB zL%(`>RnXuTMLHc|OoRPwwjl6=hSQgyLtRk3D|H+}@EmwTfI1G95BeL2m;1xEvGlkR z!d7+Pr}v+Xm?4T&d74Iwx&6Gh*ATPpl?8T@(xjJokUNQx=|4 zo1le2=;t8jBkPvgk$}ck|&`LbvaRb`g+acgBSq+QqC^{ZRBSim$Qs z7TDVR(O!MXyac18BLxeRc%}%W(4yXunKDpS14zST4dp*N5)X&d7od6J@6Wg`*n@bv> z+vvlBAEl^mp2-A4h=ck&!PmO${Q=RoF%4$5QQ&5D@l3E*kS7vY=&avAxbvkb5d!J%$n0!ru76&LPUl#@66Z7oE5&GEoGCElDIK_0DYhL#wzgm zxBa5ew#=AfzO{V)GU~S7eSKtXqn}Hh%;xBv=75ouJc{i*rmQ)62TDvftU{RnJ36b^ z1@pDl;tSE)4GYM=tZQ>z&&Dgh`LLC|;w)>+dqK+2sp2qoxt}rxW@QFGtz~wrw(~^o zBt$FPe{}U8ZmSl;Z|ho>T6{9oD=*3~mpv^%-27q z7xE^X2sBp@TUBY}Ia2|I)9e{!5shuyYB%ayI8*Z3a6%YdKW#x~tX8-?9Q%01c=6F) zKEsVloI%Xcmyd+s$9_P8IkGxMA%1wB&T(J4}>rI$zqwNLj0dE^2VT@5(=r$6bQVsa65?#CT3uQ2<$xj{?ka z)h5Ke9WyPi8QLjBxEpHcVlI2Nu&0t%yFHT(Y&URBEDx6$HF(2Rhc)LWy!&OS-p)A| zeT<1Q9{s~h1f&tTplsJ4=eisvcNLyc-z!_CHQ;gT36O+qJ2%wH(lpPk%WTmF}Hk&UE3h4vWyxcNM61;{(pTP_mlnE zMK|dN;iWY<>SuDt%&{@1+!3i*qfg3j2iidG{o-mUKk-Asin)a^cLQ;UCb=1|9A5f9 zwXWZ?^~D0<x@bhAD=Y9#r^OQ{l%#qQSrzm%49?GqqGUy~`xF!x80XD~nOWklI)lEIP zH>?$qj2V(m3+i%9E%x;wg7g*xnm?-D;L%4Jy0N!1x-Agp672*X54c)cL67^J)4Mhc z0vs_RU=chnNaQO?f;PiN-Gtti6`4ZH-PZa*|8=QR&#O#GKMxF`R+4@oZ+(nfW?%G% zgRc^=-4^9I%l1K#2N+~pNFoA;WAD)C2GNFQZ^q^F>!w{#*zHPw3o>9+!Fq%267TX=&fh_ zo^}dmksdPVo%YgIC}SIl(M*l-Mm+-}En5OSMEyK|>rE3&B)8QIZ%kbIS~t}8vE8)< zbB9L)DYxB4$w=-1PoPHlYR01=2QZnec(Yh6mlVq|;%niHeEouAuXG|Q`>4b3pdFll z+(KfPi&z|md-kkQqm8o|vfvleIVS%R5B}34q)btEEnsHAcYuA*8`y(^UjPp5kOJ*N zv7E=q0<`8!FLume|qSAFk{ z%Bl54YeU>m8W|XAE8pSZVig(M5|v2p+Y%yqoI!dr55L%1lNMtKbgC~8rlmU)J>v#+ zFM;hDjSk3X#L$w31-BzXEt8kjEUgPWDTp0QeMzu~KHcT%w~r3WM%?kXR1$W$jX{Ce zkL|q-1>D;cgf65P{KiUnY7a$Wjt?Op82gALw`{6OT864aSUr@l&R^9(CzF;10OpNM zY}1!T%TK70ZDwW0cN6^rE;RKQ{)0IJVA@4l*yxk-m8ijWM?o2uHwaS3tMDD$Cn7dv zgNM_{;7~H7R47^OXk()RRW&zc1NLq-K7jlM=Tr`Qj89x=gJBe>Tl!OP1{&j4X$i)K zT^qU*&-9y1`s$1px;Xrb3eiDV9y0E}hnsAJLSS!8Kq=i=WYId=e-b~Yc!}(sG{m9KPZUpCaQDaMzK z+}}ZSI)n~S!>^K}GLvZQj3P0e1gm?(Fqj=^ANRlOVGo{Gn^}un@uN^<84VzPCMVma zZR{-st(O*WHc?1a)#?Xfacol8@jDaB5oMOorkFZuiaR~oQ!o5Q?ui@x#$Q>KRhMFL zolYW>jX#$-VgcCOv=`vN*U}F)(-p(7`5|B<99ct<-)@YLn4$<8lDq~NsumeRNxhjt z0MJbRs`(uVh|uEU<~4*DV%hREAFI4|X@HysmFXyd(GSc%RZybKom6C4g;J(apugvr zxEq0(e@j5LAq(Mm&6ap=I?Mf4UB`0nSXDx24!4l9@*q(FRQa%=mQEFrMW!Nyq)uXE zl+q?YmL}Ay_(6I^wqhHqO-3?pCb*GW)S#A)ny&z$hXx;OwSp`!&;7<1A zj1gh9`}ITaU^(_bBvj`GQO9SIRMD{3X}@MId|p@x{IP7Y$86yN7Xs6IA{6fj*?6>A z(GF@(FU47fIoMeAnK3_|q~vWi740o;1GkL#sS!6aDX#e}uu7+oG^LqWy+37=A)w*T zh{R8~z_o}VJ%>rU?7*Et2w&Spa1%ekc&mYaRcuS^fE^h~fXdysiw5-h({YcyK& zC%NSu=Q?1H9TU3M!MmBT;6FXNX-IvW-SSZz$Mk)2tf|2r&So=t!U|P28-s1!X4<{h zG%nzNd>BavC10JP>d1(`8$9lwA7$CZ;mBiQuL&}Vq706VioVidGM8#oymKTF2s1eAQOg$P;tl8_bLo*rF>*Re zH84Spuz@rg>;qsG%~H%t6VI@07+xF>5B3S%Wcvm~3w))t>hE_$18l zyy7!%eL|xE3SICDhl2U3ISUGfIK<-pO1M|%*5;X^8A*T|0p%dE-Rt!z9Reh<=^hZv zWaJBK=f~&#WDNz2IbR{uZIAS&rIEC&w?4jSqcX^)eW(Q{DJ2iC*kuvE5n;J1;Dzq% zt~2{aPJ9bjZ1o`DX!8xY@Zx!>H0lu-dNzY{{)5mJ@kFLS7MeF&w5gDX`o@$|XUMUX zkX6Xwxwko08^iD5r0EzBXXqUFUumtThB9qO-Q9O{b$L^12-05X0m=|~p<&VV%MRTR z5+&&||K0V^ST1h$+cX(*UL@jmpR<}1FoJ@$4pnV$D=zs&0Uv;`Klx2C&NYLZx|`55K)#mAH}B-)BPpxCyBxkPEECoFWxh}J@~%d%AS`> zZRoq=4zrF$h5V!cKq7qeeuFhdTZkdzKqTVfDHr?byhVben#sr}AnY4W;$6newK<-B z06qLDHlzW72}6OUGEYsKHTC#qkn&hjrX@C*4EdiA_h4D`BPACVAG`?kEM${CwG;T= zYafzoeyw75r&b|l9PXYU5w_qtQHerBx1*E@_j@XtSCQ< z1A8R@G|>W+4=Ov$L!}W^mG~>>pXTwqrVo8bb@0UaAwxG&!$U4n+MDncy^zy~wuC8j zh0QF60{I0T4tsD_#-DvPOB38QOjK?-$9_6jV`304#$tF)0g^S#8WJ8jvpqvcrwD*r zAHjAYEFyf!Kib{;BMNHb$4)8_QEa?%lpsM+%49KrX4@qh>Jpy41cQ??;>pE14!&MeVEICgF< zB>D=EJ2b?Ev}=U0_9+^Mq-MkkX*ca2 zNDxZ(rBR87ynZ=3THjU&2Y3ZHBF(LBhX(_f_^F4)l2y7-5$~Lt7z^IR^e;B3b_$L~ zv7%jIAmRoX%k>of%iH`StF;8Fr8_oYJUSAD34Thcj|BMG;S2&0astChEaxkdB zrM-q|il7b(GtXZm=k~1N7A5jG=i8b=OQSMZ07pQ$ziK^v)rC2MzZ%u;#r;;8kFb00 zKJmL2iB7t8Y}?o~LJbbBM}vy29EMvVbYN)K8qx7l^+u&w*HX(qf*Uzfcq<&Fw@Yf7 zG_Oh2Jjx!9Azwu-z4Jewf;ASy!BV7t0aBKek%2kdG$)JJah-mLjN<6b{Xi~ zso%d~uM)c=XxHS?ejmeK51>!?DdPW+(ZW;1Tl)4TfcxHiG_J#dAZ@j?)AP)SKwYBy zCSo1}dm0{Ekc4b*BMjW(AmLA(GW_a8PWm|;o^D$^>&A0le3`a0?%30r#opXl3Xeqe z z2Eh$A6B3Rbfi}58Mi`Wabn25UMTdbk$9I+b*^LIyP!83}cDAGOwwlng;^0JHS*)9* z;F5(u+=L_%X0*Y#KlT%|g3wXsV8L@AF3k8%iic(w!TS%VQ}|w4%yNtRG3g zQGaGAT|;li7BH74l!d-iU(_Zq6@CC`0w<=%_PYd_kRi%J@i75RxaVmAc1p&A<83tp z+@blozfd#b#m9&q5j-LrPgjdCOZybd4t+M>-8VM1fgCO78OChu))^!<*Hh#O%J~bX zPb$E?yS*6Y2Z==m1G5$Lycmq2f#Kn3yZSViwvbDhtNJr4qZE`yan=T@T1<8VLu5vT z(B62#_?TiCmA4FXrAo(`MaD`)b{2d|4T8Q}hlk+Ro>)yp*{gI%fgSJ?u`kc%V`(YE@6i-~Y9m2jzA0P&?fKe|?~!3ythkpIFtdDghDIqH%0~?9xg2K3d>S2=uSw$49sr@!(x$ ze$2fAD<&y6g>yi|U?NBBDJsrrUUUgx`Id zR?ivIefXhfMaoE>S#|toZOnXpFffvY{hPqnS4=v33ZWzK(Zo(!L#oGOI`0;! zaCc?2Icgu6TB8O3`c#CY_f?-U^XBw}%1l&xBf$Gm8B;Yrj$iYmcx(LmxXT(OiokuO zXlSkre6HiD+7(DSkskPDj{qp1pM8i#aL?j&F<0)EQ)xCmL9uK8S(cAUH3b;)xg$o- zr_=LP_9H7pG9S0R1SQM0r?gBu5n3tLEL(gIpDR^(w8Zz*k3p2UWIns_3r9n^pZ^zYurQP#a6&DE396EtZ_L;3TA7G(Vq+Qi-SSX#gEXUT3)@s{ zG+0>5RtXuJ3`BgWW-Rr=Nhx{B#T5??H|htN@&OvyJ3>oE+DiPCYxMyt{)$qkqhNF0 z4P_i~SY^re0OJ{{+4ao?Z;_mIQ3b(o-bzx}IY{X9r)LdHDA zNg%nb_6u@EBG3Gt^!HX_Cw!xRQfB6+%2>I|TNfMH?`gl`lDUy7<6N6>I@=xbRT1wsf)=|C^)zgRoFBoxjC$tP%>p4nGNeu|7Fy z=#aM1*sdn*P2(A@lL5n`kl;;R?(r*1?UI8uUQ3*Q(WCykc86B*BA4ls4Z!(OhC; zZeG)$8%2lCov^%xo!|9?e9`aX^`BV<5X~qwHVc3-z8zA=$wO9r z%x?woscg;wHe1h(-@B;k8)Y5W2$b*I)LPIvTu)4Upm|;eAsHt!)5aYhzHkw*soC7H z!T4G2y`{5&ZtCV;J4J2Fc?)yO#y<>krU^}LyWfIRS>zSh=EAcP@M!iv@=Xy?2{4E5 z`FAKjgPypD?d>;;$TNb)d%zjliipWShNUK)pZZTws_n*aaj-cTnKZ2;BO}>6JxN(y z2keQ%TkXm3R44vxD0Qp?bG~sTZ^n-riH;FrM zPolv%CkyNv-~aS@I@iL;mIX_(#hVh^z%K1=#9^q5ps<`ZWD4Rhyo?@~S#P z9PPadMwd|{izw0Gvnx83xQw|^ERS^j6)l!R?3)M{zXdiTZ2 zGA)D=6%*%0ApK#C+r3S9H}S8KN&}PgpKma$aEr?Y4=bpb(v8hQ|JZ1jEF`A&Vo)6CR`3dUZO8#ZUa?y7_sTXz|q6bVp;%1;^K40Z{N>kT%*BhUbc z`<+*b7pYfaHRKvdq=(z&6q{2ny+cS7DJQgF0N}Iy0?EFfhN8d`<{e&;c=r3FdZwG< znnzbb6sJbhx4*S41Lk_cv!DRy0w4=G`{%ZoSxGn?=ok99 z-53+ss@BH#hrNcWOj3U+TOVb=v~Q4YpvN#+0*^g+*9LFU_Jsu*droJ-D}G4ErQbjx?63zXX1^a>aae*c0)bs4sc9wklX2 zbTxx?pkjDSLa81fZ%wo~m=6?5xAQ?mCPzYA+7m5N!c-$Y{qrTP!Xfh$eu*HURDRF& zgwJv-J5sN1|2zP7OoreR8t(c0VazJRrnwWtyn=9j81#s+alAZh&|`b^!qb6!tHjsB z_+7j8vw$$fd%yy;BporXJy}~SQqM*0Q=+TfaO5q!`+Ow)nOV4(d_$H+?t>BRIBAHK zGG~S9k>XJux@ku)vQP#HB^HGvsf6zX5yP{~k_;)~d!YHr zIFoMYStzw>beT8_uFGgR>Oz$RG5OV;#?WT)W}v=>G~I~+VVCu6{i!leFATFyNO9%X z3a_URdlZ11T-?a~hMD67oxV4`hto(X!H6Ps8Yr`J{6rU1Qq#7vx8B4@jV~p?eS}gDUl^2Ive=){S5nNpXHMs$bn5!G38gAEHnv$$70w+KxZ5~ zzn7{AQea_tmID-g8}TG2Cmad|_?-+9p2;31fd(c-+mLFBeEs=d&Q(NgCG}w}6gHth zKx0$Wup)8C!A-$*j7+kX+~h2gM{=M|9Y8}ew6)ah1&Suwn)4h7HeYUNcWF8;y#2C! zg-JRHAM(K;QFviM5MGB!*+Qme=;Xt)_DD*`~^td#Gqp*lby=TC19T zOFf_V8Vj?2n>YbN!%~q0chwl|Q+=ous+HEU(d6jHKe}ew^T1{%&_;N?87d+Z1f^3e zyUxQ^del&g9YQcQ`67~)6YcW_%TwDEKI$|>Swd@A7fg+TD_DQHdT1F&W&=|tCb)7! zuK*XTp2GBz?k`p`Yl8#kx)S0ZHUUr|0BOD5YS`eQ|83;fG^qi^1D7n;EdwO%39h<2 z6~s>(KdWoGm&JM=5xR-`dt}p_9=cE|AbcnnClHfn=g~=Qu96^=)X+{-S8YRA(h<9G3>U{usjI8?| zMG%5sx9QT{h6ZZwP|gTKFw)BML+)X#>N|;52_0?0qkcW!A;)F!BcFZ9FVwoa+0|Z3 zxL@@dHiLhu{3X@{KqCjv*d_kyQE|KRS?4D*`0T`3szqdaNjgKlTOMyMNb^O{2--qj zS|Y9+ET!ztpJPQI*Kzy1&3_V+UCRX@E9!A6S|q(Z5(jq_s7H-Bf&-2Hm8qBo@fl<` zNO#&YGDV&u2`y+(5)=Z}$~ltAs3WdiUQ1V66XSg136avvzn6a8L%{4afur~kS1I04 zSmpD?h12?6UAA$&C`o>mtfF(Wq0SPLwgnW4<|&0=Fpo%cWX$d@EmDP-)pazFqNcLX z>(Q)4%-#gId|@0z|Em*&xFn7uLCceaLsLvTqb#u1D)gVzR1c=%oSDI2;zD?vg-rT~ z?BYa*s{CQ21*a`!<-;3v`awKL#ih|BzR+g4G%|7YY0vq;(6tHFp(J2`X*LFWz(WCl zd53TsjwGBWCWN=-zWFuHS8b>DEIfwdL4}{R#hMOk@c%cmt-@UzV3K~XZS2<dk zk2%^+Pie>Kd@akY*Tlm<&#R>_0B*z?@^CmGj_mwXQ}9IWbL(bZaK+ZB`isEwi~T@k zgJ$|d^GLicaxc;}@HLd$ury8YL^|_>BGUTllbwSLw5a@v{rdm>0peM1t0_zXKkD+- z7+GhcgsIfa6fLAmx4<#dd59i7#FMunu}EQ#%xSD^y(k{*L3{J3mdueaYxX+k2H4E` zGpJ45Ct?ga;R+2>rg>pSj_x%zyUVGr$@7rC=M2m2Bli-?(&y3TJ$o$DBSrd9X+B^s zAEJ-xwQRY+7cO@Mv-I>wJY@u4R@`R>ajoZ5e9Wlb1=Olst&?M7Qc;u(+ovi7hv_sH z#jwXU`KqoPY0ht#uPChkoF_T(M@4x!+Re?i59AUtsy|NU7bPj?+bQkA8PNJmE+zyZ}aXgQcPz&C;_ z87C_3h^#fkbczcj64$-`f}7`Tqs6;S-*Ki#QL`upB&yV-*3Rn3SU_IM8uzhDV}tS8 z0n|*V=+7wP+#8=|gYw@+q7H76w#3=qzeo(wY*~&Y8I2pR1X0!5ZM3NMo{G;-#BjS7 z-_A=Pq2Nt4zGPL1f*TsaVP@|}Rxm7c6Geo+%v;}JN<$gZnQdjgRAW;Bx{8V8I?(tA zswH}Fn2(>~-dI3b0F7B@M%Z3DB!#I=KtX=~L&NR8>EuHpMDZpT;>2JA;v(CiRIxHE zCVwFjfh!$?p-XG!dPlxf@_bN8-bMXR>P^iRXlmUuq3Lf|DE>%c*m;KsASsFR&m-LT z?k{No^PCC2`1ew-zfm*9TXqbzMI*#Yml?2Hm)l6vkx@|vb0OO$4U4)_+%$NInu&^m zR(}#4NioB>+&Mt0^`7ceD6(d9c@iPTuMJzRwMTh`Ep z#hCD_-}6#pb&Xi((%kzSfFwKRm9yr{jQcI79%R(w`I|wS;>zhsk1l!X&9&1g324zZ z?gV{e8tb=3pIXrJb$rfxbIPJ~4EdQYpK49H#nkpIy@m|%r^c;%K`_99tCD6G zT+nHvPd)bE!HgIk*guhR4scD13Fx##yMu_P$A0_c3S7J^GM%QR$2PniH@tX_^Xd+L zp^z$W7!t}thKUPw%1Mm|t3O>x=CTIDS@w3RB#yU_&7c~@YN2FVz@uz|y*G3^CrBJ; zKzIb3s)p^=c&)-?rx&|&#M{;7Pmyy6SCd!1sh z8j{9}>JuXG0R@fJ7_*+I!k5JHj@W7XwK6mXAP0ad! z4)XAee{2y~Pa5}*yPv!{;HunKQTG-GQs+K1WtB}k^qulMU?S@p>A$bD#p9t}La54i zm&W82yRS5@7jwutpm0|x;U(r>^R;Vc=8YU;BluZw3y@3w*zTD~e1u>hpJU69pck+V znQ99jYxF{ zb_7)6e}YtpTa>-)h?lwk5J4>470)}KG2__-T`VM(pCIRfYA>Zt~YbJ3!9 zi@{hwV+@ph4D2G%O7YdjWY(mTqhQth7TZ?PH%vOsbS9tpElbzY9K2x6+kLd`6xc=|jhn=!&wC-$5GmmLe1rL%ajU%(5F$pa>ejz11|vga4&-Yr||@J!EiooUXJRsQ!KTb1F~$ZFZ9 z4qXpy^_Xo9AUc+nQgVF*`KBv&b0m|gD`iUn3q=3OzqkUrvx2PCGvT3MzzT;c3@9#B z#zK~Bd>X}c6JG^XzSXi3%m3uqi6A0d_P46*D$5+tyw7+{|FvtxL$LD052BNDvo-@X z3BbG=YHXRe60qS%j`+pSJ(hBB_)3};VsfB07;H`%1Kw?_9g<@+6=dG3KVlfcR+@B^ z{7;+($Jh=~t=4A zKc2DtIl?w{c->?ekTMYu1&*|EbJ6SgQrcvT!}^tF=243%4fgja%#4X4=$<)p)cJz+ zuyUK>u4n~Qt6!>(vr`_?WoONTodx}J=zT_ku$zkBMicJFYTixEU-%Z;+DMe4UHocRx z#lJ#Q4eLu`BGan=|9J9OtCw*<5>A7#G+ReykES%#mdE4q369bbq%_ZiAWggDRP%d0 z#}x*_ka;VxN%P1I!nr4|nwo`J43#?7dVmA@>2S>|!*0H(T9NC(u_rsxSWLAi&1 z1YmMF4`cETdn3WrFk)Q%6mB&VcgE=voeDt#xun5;fnBDoFEqN}jP(Or!MfuaZy(P$ zB~||0d#)cx8>?dF>sxZkZ%;O$AmF7qp(lQ|kcKa%oi=8!#~D?WD%gU`)y7>aA!od? z)V(;2l0tq_#VVm0R6pF6+#sV>vM`a~a1x_+^RH(&2(M9koW-xgC`WoYH-&M{@2&LI(iPW|*( zG#C(jS+o4Vh?AeIyem3>sKoGvB`#V0zE^X*Yq6Ue@Js>-eX4kcC$U;^rQ|Dpgy6to z$}V*El8&0T0ys;azsG|X3&ImU$kp@v7|rkP(fNu0%O=*N9U=Nqy+Twszl`c5fu(1l z*(g#0O}t4b@qU*{_71y$39AY&wYe~^b{f@T{0>sg^rD6P@~@T^s#n6hM%dk&L(^CS zO5Q)o;1I-*XK`?rh?#QQE~WYuGN3vGJ`I;Ywr|#jI|sWhY=zB zHd>Z5Y#NFrzUG=TqR=^K{wny5A-eF+-SG#1KCjV?*U<+mnU9Bm*vg~v1dP3(=*rzS158&d~6pn zxMrxH?E-%USC|YP8_8E$|Hip{m4lsLRc;&Wx^Zb?Ri+46WX9P)fi7G&E;iRUpzZ(` z+tTOw<{J(7v!C~6w^T#3!mdh0?5i3af=kpS*@(~hG%qIUe(vg|)A!&e@{Y_r%-xz6 z=7PighZ@3-NmWG4o%-mdi;CR;awd_#=M4X)yLEi`{%;nU!~|BXSLcwLg(icX6|h!p z6!HH}kM36u5|~e-+b}dP6uaSZ0a>%tYsh#XOpzQdJ4e@{9=B*J0nT8yNj|Kdd)tX_7i34)jv)l>k$)yPDg7qIY_!J#rfp4qLo?;okN)M=Y65Lq$dt|`qwNm zLhyhDUTc$jtZ zMRC$fuaH#oz`}RfY#K;b=xE1Ec#{A_lRYHpsj--eGS5vD0D=luFwZ9-JU;3!cdw=GwA#f?vL((F zTbe5>yt^PGLqKjvA1=pBA;FbR?(iK(uUAk!u_`#XRa&BlD@z4;Tt@~DEWGFvByQ{P zEX87hR-y_f=&x!OVxa8vR3PP)qOq1L}2zoDH&oHG=Gs0riyS&x+ftAwsP`f@*PvU+!`s0g~`?0wx1d=M^#gI6t%=Av~%AvA;c?5nCN(Gf`CAqJN&eY=te z);*|JBd4(olS5aT6QWhZUeINtBC`gVsgc*K4?+&>Gxzu?M<-!!@oZ-NRxPEI{pXtJ zEHQ^S7WF^?lv%*IM>$L*1};74g_Bj6k6J(o6FHH)?cnVU$Xxq{L17_n+uSx@a^)Js zV-f6%9U3wV+{)dp4M7@!ed;IIyF^HKSz1vh8i{2YA&-FLfG{v(JK3$l)mM>OQC@D| zh~Kd0kRc*GVk0M0)GcbqWQ_gMGnwk>)U50c25hUSRX{_&#=Y{?t&U=UtR>n5fNYcV z5&K^CM;a)2bk$Qec@dwwSr%&FCAj>uM@V(^;p%QhHke;D$hG(HGuB19B)d5esiRrl z_nh-4aQ?G1;?J8Sok-)ia)>{(?>bD&)Ps-*)vb4kq_lPgccxt{(eEA4ZbouGZM({a zw$%|ndgTO~U7=nr^DK=TEB0|5q9$p3mp#QrRXyb<(sw=P9>lgZLvLiS&OZqJ=d0KQ z%9#)fNTlLa>e3R1VvPiT%2(&}>1a6scN2qiTeOer+?Sfm0a6j!HA*4y-?D@u3KV3)KSA3Y*)D3%!UN!S^C+ zdR{hP(roLaWy*{+#NMvSnw6w_EepQb=5ObY6s$FNLNW;6RxO2$jzS+_ToRD-M8eCZ z(t#8VHo2m$&Xn&r1~Os{k~t;bOcuBK#l+a*hm5vlHy7~?PecM}^;qYZ!wQF5W60QT z>IfeN^|swL#B@izkG__`K#k(>k^FHE#jdrjrITf#NV6~Y_KEPRjcga^bFrx(H#i$y z?KQmJ&%%BWnZP5Tab-gZ8SAj}aiP+|c~ODGr%cUBHate*Uv8mRJW{q^$;V)Ff&H(oF! zarS}>`RF|(Hh|sh)0U1dAAUH!;!e$LZC9l2qP?>WR;;G%0B%tuH{n9Y|J!?{HC@fJ z=;It$L`YzplD@Om-W!^N4V)@H!(aIjRNq&Mtx2E3703=S_2M?OZZ)L-($U>azgUC? zS%zqB=X~Og+LXUE1SnGO^huF8328)*J!(p}hQCejWMQ8)oG5sKMR+|v>rJ?v%CRTa zDCD7}?`D;jT_G@uFw}_#5?IwRo>&_M=YGE(c<+>+CVyQDYQR%*D&F+&*}DQj3kplt zZn8b#9j8Y-KtTu;^^D$&T!^^At=HTaJ)WfZ_*;H^b-Pic(Lo2J3i!)Y;O$lbN{7lx zT*q_&^SXz6Dw^SV1W|_IXVj8*cWcR8HJ-(Y3z>OeZua>At=(;cEX`qBAq3OvfOoG> z$QJBk*VQ>$7HLrq%lfY=-83e1;hyAx`=;DGCjB;a>5KibYWD)O2)*GkyDs%S;_Uq6Y;zN(G)j`H35_&KLS4%^bT0 z+aH-=Hj$MMf~Fh|wp9M`Zj*d;Kgs^M6-81d*@WQ`>}LB53tG32k8}OuM&baFh8T>T zah%^L+;&;y3Ixl1a$*#~Ci9+4XpFKkh3{U(9E-?Q@OJPa0OvJcYp}N9xiWx@jh$?LXE>i_`zhW2Dg63qVS=bQc35M488^-OD zRW^YlymYC;TB)@xcrT_f_r@IG2W^xsjd*8dIl1LEB}JhUnok)n{V{9y zl+%bW4Ef^cKRe5{m)+#Va~*0rb-Zu=R`t|d`|EkUAty=Gs}?AqX)&v-lu~rucJI#U zY7E?RP3yq+E&p#O`m9NKU$&IDF(hJ|a$%JZp8Ux&zCn=f#P9gmq5MJjq0vxZ*)|2&pVcNMIeP15Ly$EC+Ycyux(0ZPH&e*YkyeDD-D}xx7I^b=6+9 z&EkwDi$0+t&Wf7hi9obmHW4GNVXk|xcj-l~1M-61)S8iq6q?H!^nGuEj>PFVI-}C# z9M=#hhcSSZuGhP>j`(iw7asv&i*e>r*6Dnf2MDn|5a@(tlh+e5`KnMg-wGmN_BD+qITBU>)f|Z4sAn zpPWMnrL|$T6mr8S@|#^@l^@Y0^4Tk;pU%8$F`6bY0f^xyJV4LVXL#ATWe;XGJU%Ye zj#IP;zdlX)Q4njuQF71yrX$wY#*8+t&$hfb#qQ&=L^c+uxB&Dgk*UNZz@m~ZBVp@^ z(4NDn>K#R->VC)yJqz3Fr$uAzu`Qz~XG8Gg__boZT0j<@{Wm&3Zp?~Y(^TK$^fi6@dy~)ME~e5f zjskLxFiKSv#kPun#xq=yU><}-Fk3yQiK+5;fr!M!Opv;%4MpP80HWs}Cgubxwug1? zIoOH=^RKx-LB@Mbij_IdbWU{CJcG&Q?>>^AQK2H{Bb}c`Rm_Di=^>(8 z;xRlQpQihadz=%^`u^UawTY7s0#gKAlp|EDv9Iw1nhVb^X!~`FtaQ_gA9iuK^DDV= z>w9JllN9SSuiw|nKanvW_RP<#tIKCU@^D8C5If&R)jF^5_^*g&$}H@+hA|8#td&k1 zSoV*EbeN#PMZTwek#i{O?%6h)-f}yj5-tFJ_0jERDAV$8V=1n{FL2^Ye^f9{!#G>Z zuC{5E;hN&$M5+937SAHjOV{09{|I1BBwe=^Z3ornTA2df8>pCpI<&YplN%X2!`u)c6vegy|81HuMpPI+nv9y!VYGLZC?1$-eOu|XyRMtw6#-A*zEghzPlXx0U&s5A)zF#e`EbnnVg$gG~7)T1X*!&a>-Tz(!(7}S)^ zK^K&ewLdYeVqnSxR~*HCSTBi}Zp$0g!LpCoR0_z*xA=5Ud*T{D5BnMci8BX_`RRhngFX z^xm>wuyIlzu}>g$j;3acl3V%mJh4EIN98g8zE{s^;wWJz9e$Fz$|jF)&Y~j=N#DD* zRNF^u(|O^ik~YH_FmUlPBLEAYsTzupz2uqeA8Wvr@gg(e=ws4{Q((}!N7D{TyEZn9 z->L_Nf#EylM{6mp9gLm;}ong-uag-Dlk`@Z7rel`xtP_qzcf^QY{i`Z}H# zoU4hiBZH!GnEF*?0$*RTZ)u~SLW_MK-ezmdAR-LWqX^>qE|@m6poBAG+)D<-cGdl? z_b|WOdtFuanjic=>6Xpn((U3a>dIM{8u_DP>R1J+-^dbg!PHB$yKFxu+p7~44!!ek z8X1;NaZCvTUlQo5SK5izF~j04Nq7aI-b>W{ZCG4R($p1qB5JhG`~H;rkiB7K0it7I z{34T~AC<%X11miqlzrFsH#|=2{Y5Df#b;2y9_JzKa`Ge+FH`GG%qLfR2F{?=?bN7X z+HqxzhL+f)Ag6j6M!{ z7zq6ddWK9QJ)tlfOTw2>Bh&)wPZ5&< zDtFe+Tj=!c>{FpA&H?t}bqSbphPI)10?;9$$`YB!uh(-{(3qM?CAa>U$U|X;`;?Fg znJf{IdZaXnLdCL-GmWY3ncfS|xsq~6PK_-)%i>_>f{!&CfaqviJ(n57cSqK+;Pm%Y z#@9O_nXFG(%=cqOp4SY|$l#=-CV*ggFP8=yt<(5)tyDEMq#DXxISLYoL!{LqM8+3x zs$#m7XNElsv%^za+PI21utzYzr{&(`w~}9KU+x(_9MAHf(g5IQCo{9B+>x~`W1*~o zbp@;$prQI-#H_tnV0xy`<54EC4;*p8Zs~qrd%;>VY%7VVG3}iwDX>&EW8JVH zbdgCPFpKlKYsMq9eGtfDMjo1jeGhLgalD#Ee41|fOqCpmpt~RT=lw0L$#f%vgGIDR z)0s=!kA1e=>6P%sLG9va8yMi-7r{Ch%2U{=_h^97t}FHVB;RXgdN8*_s64{-aThVLP?ZYFE#&tKVj|1|ieV9yU4Lj`y_aFa%oMOlhmbdQ|gx3j=%+>H`PBp0vBF0;>y~ zjLei)ujq;0kb1DNdu_jhqO2tuF*#+z#$B^*0uc_F9Ku$&D6MPd6x-rg6$RTDQ&1k* zH?hh7qi>U*UgCE&F3gLhRDqQ)!=>_ef<&=Jm=OBOcfAAJQfctH$2%P=(c*u zQU%8X5K?@oxe!#d15H|+*J4;0(-fKNiELJr@USYW6)9$!tZpTjbeu{Tvuy=8^f4C? zGFS2*5YJf{rES9nsr3-7r#w}#ch+~uOn#bP*-;**-o2W-1m|@;BAq~87;ANA#7Q5y z=CV3E(RipBvceqUJyy`%3KX+2i^dFC`hlc`>#+p*Bc=08V4TyErfrQI3^3ggRV%SE zU8?hk6nukicbsVTp5m?jvn$RI-iYsO-KOS1m>BPSk_g&0rjeR z7Wq<5tKl)swG!Z_S%6Fcv#141cH}6RX{P+QaY?n{@X^L%e~9#DPPdAN9TIOcRJ%kG9>1?1f0AN z&8Rc+w@Fr*0eCazNG$$SFr2m7(PaRl37bP|o^H47x>_%2j&Wb7p(}e*Qp_u}#jC^j z)8(v|2t#TXO(1FH?j=yOS6U5XUAKGvhqL~V;oki!s~*9)F5a^>Nwb07gjfscXP?9ghujuzRRl_9@P#T-}I5WWIRc3v!~LrR>O|Llx^t zFn<8q0NA`Pv44i#v2k>7@&WO@#C+#E5kwx9bPB8AOcto*HD$mHEYx!C5Ep?!O5n=eJNs1p+n|}xE+rm=niB&>l z*mzDxRuZkckq^}uCbI#-=?Zc#g6J>FJYcQq$CSU$VCbIMi6NAAhltAY-KRHZJt3ZEes+*T@$ld8=qO zNnDRB6)kG{au^wo_EgCf5-zaK7;Egj8u5XED!?0f`NKr&Vw;p4?AJ{fSx;(<>i0OO z_~DC@^0^L~^+&#}m!x=7=PXJpDQzraCs~76z~0hs71;l@es?j1k7=vOPqYFz2C-b% zDbAJl#TNfkR1P@Zt}AQ1F_(sMv_JM<&M1?NoaIhNPoG|FP|Kd{Y0Oy-u_e0Web6H} zp{H^0Eo4K~f!vp+I%51{X92LoR})T!r8RyZ6p~-OpiD0jns$7 zN$95hbIV`=5j^(#*B$Np#w+_KJ3yWylYn9;+ahfqa^-OyB4U>Hl@~?@GPj@L_J)rE z4$V!H{KcR0O59w+7*`Na(+4|9B2M}!OI~jy3-ua~@UY4;Eg#ARfT%A0daM@5t!VIj z2UAO9ki_AJw_PjQ)K|%gRevhVpBw23u=%M1{VglokhFIBrb$H*^Ic zq|oCg;HYTZbKt2@ao{ZCQireSO=d;U&r}@wtWS17=!PaDMurg!Z5ivZIC@f7DtKzc%of~o zx1woRg$i-!mCcpYz*7;PJs=(MOjkFMx8CbJr95GM#Wh>IMsN+!i#)mOW}j>5=q$r{ z!P!0%s8!;)aSdH-K4wv6>>@yQ$nmu}aU4x*i>0wiez5PT@4BztQ_-(1X{~D@tPV)_ z;GWdOf9vqgSAG~nIZ|6v^PYtX`^9Rr22ba@RNR$VdP2edJnH$m`%(;nib^?Z%_S5j3AJ`;>pn>OAvtO~7zAt*{k zEp4y1QH!DQQLpw{^cR{KqZ!4tK}_&aTN2@(VUVtu!litmK7;h_URa&W6(o%hlSu^Y zxrY|UeF|M5OtA=7Yf)LjN|=zjY5N53tgHl$4ZBsy>8JNcvTyl^V76`qmq{+pQX{?( zSF4mM_?tzxmCJyp2jcv&P5#)>^lBrS3X>d*FjgGs)-k2JS|L9sdh{$`6{@WO2T(`N zuiXLuhI!#!pktbwMfucP_gH&f+*&wWak)3?bkNKacUEt?_q=wJ`!eRQ zM_*U;oOusa8phy^gL8WfLbPvD{^E9r!2e`n14WW=0KfruR=qV1bY49?T@!lyS4D-l zBbD#w6RirWzabbEo5|{(3dDWVI@;!l+KX;%q>+|ay5*U_TVn@U-jt4m3PF5=wF@%> zc>QnQ+iiCeGj!5BNIDkIY9hY{kDNtr6CRc&as8l-pW z$(|A{vf`DEzr9tiPM8O?iCK5i2F&PrKIMGl&x0X(T?GwUJ^jIJJP(H>h|`|SVNPYe zHKl-582nyBP0}Yn$kRtP>86|mFAK|CBL7i#X;#vFuY0+&BxG zNj!)!R^{u&OCLXu)MVifM%OWUVsE1B^Q&IH z3-|jkwe#8#hYi$i$J?ruc`RHk>McgEY+*r&Cs3hsXm9^{0OXrYY2||TW9{fP4aZA1 z9$0Nv&0zijeQyU?P!3SrXszZGDC}*HHWrXXhnV_qG{*0bISx9h`QkXJRaN#{9P{4(%b1st zl??mx$|rUPPKNX9K*u=yZq{`6ev={qsJ*GPmhIX(8)6!ah58HS%{<%vWkYlPtipW1n+4i(+U|tI%XK=rjiWF5 z`*}+;Fc>q8<{`I&XZoaa2P7sCbzL9YQa`hCBD@EM>Qp=7=*;vn2 zA?R$jAJ!%R82u)5ftxkmLFRElTk>x+kk|a;?0zPjZ&h#(vcC`wioK64gS>u~Dy9BQ z`?se@y$VR~K+8)QQ$K8h|eOf}9(~?`xLuBUM91BKaZr7EZnjL&weY+^70pP$P=qvi;8(}+v)bsM35&Six$T%(~%^MG#F9V z_-Qzy6dH}Kv)etjxTJ&up!3**GmvtZc!1`)BSz6^Z>c*%GHs(@f$DQlkMtz#jtMB9 z7ouKq3tT-J=lm4qGE3qw)SIg$IuHWXu^9|s`&5#|^_?$M$jObA)G3D&)qec-&*ja+ z?Od1O+m$QW4jzak9$LUn+^bP9%kJjQ4r22agZkKj#>NWe1EkMQE>JS-d9esVGjg>@ z1}Qpc`b(^bBGmL7E6zQG)q97MOP#p}iRL-{=Qv#D(5V9?YpB(>>qpG~p+`|voXpZy zkh&+q_A|%A*=)oH*iv*72vtJwhi6vPdn!1rV1ov5+R0$a>$&R#Y=JrnP36BYE{Ei7^UtU3 z{x!rd(*Gsfy=S5t4g4q|k=LGvL7a|V@CAi@c-yH11EK;f(7}o)*OBSV7^F;viap18 z=mO}Elf_gTBlAPiIlC5@Bt)vrW6}}1oKuyJwMmF6!0QF+^S9{LW0<|IPX)uRQvG&nEwKhKHqVgq89} zPVf0pV!l_4#VJ3>Z@>c;FPeQ zkcxG3@QC;QmT$_`v!zm%6Ah|ILG@wNL0NvtV4RvEWR@7_rU>Hxb##wGE#9zsQWFyL z@fSMDePmq#?LS>N$&wfy#Fmb`OzBw9BOYSwGYkuZJVp@G5@sJNu4sAIGxzSQxK|KP zqB|z0AP+I`@Tw9Uy-Sfd?)o+Ewwq9v!|>oAFzMXFP(Bx^NI-?IY6P};R*~Q>7bc|}zz$~Bi6tUl1Ns{`s5;rG-8rg0sfn{SBSVnppuWigZF(vYPuE<$AdFcpdmu+=0 zn&F@5?;WwL1(?cnQxg*l&ju?5c{^&MQ>EK;a`Hti;zM<`1Fc{K=PYl=Vc?4H_+Q4r zFnICT!6;9(0n6Ct-Kyh~x3R!!B?u>Z^9(^5v*t>lA6yUp5q}FLN#gU2NM?(%HJ z0(NfKan}=`$HRM6k;6uu;76T`7Fx<%w+ee)dn=d3 zOG^&p{|}pbzM*s3Wbmu1!<~vxjgG!Fmg0qq!O6uYL9-qXTVuhmHwmW4Cx zS*b50tW?`Eiw;Y8MUvr5aF#5=8{lWuIIIm%LA>+gn}v*@)*b;AlyFnwU@M7>FrjDP z#B~#+Ff{RD6Gc!2u??qa<46+4fcMHsfoM9qeaHv+T-7>}1|ku|pJPi&jKcR{lQ!bB z&+pBGtLWvkVkNKtf=fhdQKSKjaoWiUdg&b4;KA(WsYwuUoi0oDL_Z*LiylN2##8;~ ze6aW0JS|}v+cRuo`8uJnb81u&GPDG9ZLW*ztx5qj;z8uYpLad#78GKOA}cB=(Ge}& zH%I7h1M$ovWRkg-{u?Mix(2lRIrvG!J4IGkG=LKr$e>J)DDtvXPfRr*4@oGj);r2q z-VoOIsn3^WR*&FIF{J7*qD#I)@@bk4--*He~K zUC^Y|+Xw02IaF+<-yAkiaH|KZb(@QZV<5ahf$3?eaAk-tWh@<^00UGaK9!L)O&5l` z$J(ndN@7j;4)z=M%emi|iLG@M==6aJ7(LXzg$qBFCn!_>R9rv~$FbghWyAc&tCs?t zg4QvOK*9y^9-}VbD{m1dMl_^5Q~lD@fZ6l%Vt}z@a0mVBf8p*S<+0*?2bMyoTk0G= z61bZFp+Wlt?~HJbkrpeY8zo23N*;W6PP?J;6ih6t@@VRt&|yCZx1+))>@gT+8pr?kli>8psHzxEWw zJp5??%>d-{*9VJuH&t=V&h!YQm&3T0Tx7){(N#VwR}Upx26+!myZJQw1F7*5#%`dd zF>1zI$c(K6e<{?rb(VBNBx_oqKXNL;>nBNw19)~JMOPDPobH!4W{>A2c^ZXM2E|mH zH%)=s9Yv}0!_<~AV~1beA6)h;#|Bn~75xhvpoFgrn|#G5NV0lNUgXi;H1Dn?CZ4S{ zO-j~4#u!P>Vy3?z2gk*q_N-n52x=uLRM&c8EoIy>76we|#sY*E0Rn!N6Z__vyF-Hd zk%jC$M$xf(S7`wQw=j;tinEWPBv^^Zz|iqE74lUuu7C!YI%g9;Al^hO6TdZ$u&u+l znLb@-t(WGu#^gTwFdJsAjHV$0DlHEas%E8t*Q4uj1Jcl%QhfT}z|>@s1*jl;0)JMS za2oKK(32lyrQB)5>d?Ve8(zY}OXHibxu5isMB#?N$EgJz)2AYb)PmF zOPCY=&3*OWG4sW5R(d^Iz@z}-8!tZ%(GdZLl=33*yF;2qCwnW0C+MO|MG+O9I-!u~ zD|0Y^gH$^%%D-8YyD2O#ilR{iBvxe38mr6B%b*$lI7$}P3)#s4?dJG%f5x&u>W#!V z#BN!}3VS5|ffhZ~6326x5mS=bfoT{|o4mt=$HOWh@A5T&rvW+Zy9$3L)~kHt0RmX< z?Yo%+<99$pj}5I%;4v;C@upf*@QIL(7E zp$urUvbq-kP5CO2a=IOSJ9A1STG}} zK|a2D;Uu!!fy3^*1PNjdMO)NT~J0a?`HrRPI{35kpq^_vFVQJo}j*mpf4@@riL;e;n~D14HsRZ z!GS3d?(^q&JW8%UIs;2{hgS2Dx;p;QnHC+)T=WK+B$GNZ&>eEoA9vx_{wv~v*SG7- z2|horYn3FYCl>B1xJ?U=7c#kXaS6p5y;Tw;YaIp6=3sGYa%VjMs{6x$xx?`9yV`y{ z8I$)Xk?Qlla(qzTeNXSAadR697{sS+dyYfo`sW`SPzXKg;ocAusdz&tSKdPIiA3I*2?aOU>5-MQ+hJ9O5 zWk+nB3f_8zIyZjwl}g~ISsfcgkj-KCQ+*mbhZvs7jCOTGy8P+z8WxG}+Z(#1lIY5{#wvCfc`Gwgt1O8Sxo@~p&+ z*Vg1eHwu3|4t4R~-S3T4{)Ql?jZ7Se*lVP37vpo$mSbN79P3InZH!I_`TD+N6Ul%O z+n8sWiLEc(lMw^Ck>m@mSV9o?%R3}$nvrZwve``^#xF6UmpC2BL3~{3aufd^IMhFT zu~_#ZAM$P>C#j*ne;OO~)sXQ%+tG%|az)^?T-|LJ%UNU9g>g6_89KVgc%&2QHYuHm z=N#onp$k?DIGjJ|igqa~}>;s#85pi51=fh=;pLy}R{lF$@ObK=UL5`(? zZFckeJK=AsG?6@*_P%O?5_Vt|XfnXH3lw|+0mtJXw)d`)ws2`t2_WVQaO zZBUOIp4gN|3;0^Z;Zyji*b8tZq{>Jb@prKb*R+Wn^z1i?`S&2h`Y~`{ zljQyn{CyNi)#~dIZ?H*%&D*|%liuJzxHVFD!2ak7-PunN@kE_ zK%l4Nkbhr#9m+g#H*b$AdDt4$Ad`H77VzyiavoNVUEGRc+0v$5JnD*hW^wbBf z`Twp{chRmoRI5R@coogT^wgJ@Co8f%z%a*666QvMukTfYir9{t4DM>AbUn2Ea6={S zj>6d`?)rcN*o(Zbqb;rhzWiP_vzblrB zs6@u!2s`d9wv10$Na+E7L1GwGN3nPy6Ka+49{ZhOFaZG=6KwT*uKTP8h_swO6Vkl= z8c7{YGJ_?-a;!k@a>M6Z2sl4S8w{`4<5QoUy_TaI13LROTj2`-UAP|Tp%8iz?QaRG zb?`%D`RHprQY#fyXsupsGr6HGKg(0a^DQ=&bRlE_94zj@RK&2lm7UeM^~Ho&+VcOY z3o}8jRJuPxL%IzdPH2Aj?qZV-;#Gzoxpp^kL2vSG3J*5eKSdm!I)ph5D!;Va;EMwq zw$<%xX2;Qq8~@?r$oV;)L$YWKXe;HrW``YisU{B0_Z?%PIW;e>K)Em=z$`=kSbRb} zqLYkd=xJ?;?N(uj!+nFVqVohZMri7R|F=Tpn>ENJH-JbY&$MmMz(L}d^xG3~h@@;q z-mqOu*u?*?&Q%S|Vvc49Y^4eI>0fR_q!EX(>yWap9d>ZH`uS-T2Qs!|w%;~Z{>d6S zRRDI;j%gJDFWd!T@m9%`N-@3M{bkh3FCDqFdx}I z2p)z~Rudct10&kQGzdC$Zwc?(ZusrV+U9kd*)KSQhdjmjt?-P`3C5n^MPw2dLCpb*OIM_`#^LU+M%;VWR z?`ERUNOYrSx$mSQysL1FEIAUPCs19CoUAxPO&=}jrP$7yQX*4B5NphQ0}<3_11eU+ zw3f64&+zHdfS$pL#F%Iy8Iy39@{-A9jUP68FEt* z>8AP1lQy{v{(WV4D~#o6u^eL4Bz|>%wO0L%)MOXk;co#;v^;eB_8Kj7Q0nCnt9Vt7 zzgl^HU9-!%<1kucc$MtxY4;^YhAx^aMMbJaF@|Enlz)FF^$?8wKUTq?%(pF%%}gsuy1oH`YJc?HNQdv_;O&KvYW z?R6VLtx=$qvjKdoyA>Gh+~{6f4245}J4$OqkEJ?HLa(?oi)$?(b%0e36S%R$5j!_K zvJxRtMfjM*5kcMSRJ@Bh2=o(84IY>%{7e|8@kz`N?_L|=zTc8am-pN~Vmn*Yf^5!- zS$p>a>*wHm;dvCNG|&9~gfEGF`D=gft9ypeqoP2vdXg@K&!Sy$chM*7kpw*elP~BPz4j6 zWyIKqv@}&0u|&z+(-TPt%B*h6o*&Fpr3Ex;!AVP!7nIx%7A=M-GVCjoR)YL6V^%9d z4ut2?@h%!Q1PhqG0DG|?>1CAif{|b{{rzGm>AAaz@9&_?FB(djo^t67F0l^Cu1~)< zKmt}|g#q7GG1lQ4`|X2IJ_8Zrik<9F9qoYfwdqo;`Nmx42El{^=%K)0*U@qJ%jZAc z5s5AOtnCS$el$xTs2?oFaD_~?Czve0Kbu%e9oK0sVJm?FmfPhBcPXq|Ha8NBT>`In zK??_JGxundYh=g;jG+j>4kI=8zWD~a#5s(H{VsbampL%#ghzD(?jkRXl8Y7Q9jbUU1Er8-EBzf7ADqRgF zssa{^Y}!#>wewiBF|1(4<@}7m-gM}}USmpaO`Cy?_WVU6n;1b`8{?k#iEY8uKy%*M zhFsD<@FNNQKS$=i!)b@m*e)G?!f~+xAWkwFm@+X8%rpj%g{nl<{6AZcJ`A^6L6fdJ zWSdt`6ca~t#s-VoUbe!Px!Yv4>Z2+@;_XrP9eo~Qte6-YjG4f*u5^%Lp6E#TfYtWg z>7w5YL5p23POl77FsK4+CcriJ%iTjK)`@TyIpjI;nmCv`BO{tHmL|jqBYMe@uro@A zmS@$AL1`lI0p5X8g>y!Zu~aUQZAjn}Fv6G9-q+o_J8y`c>~J+`h_h6Ge#m1R^s+5u z`H)p*FXo%@X<5;mRX4Q{AHhx5NAXkN|2I5cNE8ZdV2&TRU&5uG`OdB12XGn;lG`CE zpDv{2O>_b&$p*X z${{&nmWy#O0Lm@x-G`HpP@8HS3+X+|N53eB0ajNbSqnvhqmrFAAQP%S^QUgxjR8nK zowCau;Z%qqMDMQ?rX9+Ov4sD(&$;efPq+rb8|A!!paP$gb1^D$7AaXHqAJ$Cd}yVBDmx%94TLe z1gw9GyOO4}2-XiCSSyT(%Vf!$CM{O*bs{lmtea|6-49rRj(pIPEP9WoUR@6lQ5}E+ zzv!QB3&*{TEQ80c;=J`c5Dm0h=jTq3r|y%WUupQqyOU#dD6Oa{+dq-vSr9cXSC zoX+_#OYmZyCLMwkRi_PH+wV_|`#22`BX>ly@I@HUNHCjYDBp6OQ=wp_Om>7 z!R@#DcS9%(gNfdGIqel*gy zK$YJ=qJ;hF%LgujX2$>LH#Eq&D$5F}sD|`9k8g+4ca4GecQ?wxLID5|)%ed(N<%S;ze8s%UT9J3Ct_Ma8RP~nk z8?s+eL?drI;|l-K3gk6Ez1-LT9ls{SUV9LDDFpWaR#p(D#pb~=2o2C=QYL=sOH7CR zJT5;4Ug?ftN$HEJ^V#F(P&q8e3i7~3a_Wm36kYKwv2&J5?U3I)(WPc#T(R6gQbVCN zmL6;bW^1+j0C{T%T0t z@8{r->vd27AGrNM&n>IKu(yP|c?25n?OqJfgJOcb&q5ys&FYrNSuu%t-uwHJT+Dl% z*SVPotS;col>`Z1FjTtxVm~Kg_U3|k!Ubsu2m63upocCTyLxAd*#4e zz?2_GwvLpB0WIt+=|o$$0Y}eYY?kC~qXu1L5qF6end~|3t4&Ss19N`J4SYI*@coGC zJ4+b#(2({>bW$A{{1}cmXgsjN7ngy_LH!zZP%__XKk+QKnW}{ee4!{%Bj7!+geFVG zrMY9$65G(2wo$@0@cB4a8s$}uy(@o8e<8%OP~6JN|B8Hq*z>FbO`Q{pH0Ly#ohR(h z?dx$*y$z>}5rrI>&1dEzk6BK6tm(9zKTA1Zcg`YIwCdMVgPW?6U{3O4Oj0UBF5}Mw zcR0hplF2wrxpn@!>nSUS>U<6tk|A^v04`}Fcu*axaSXX8H_V}_HsoLFbh3w`i#>3*UwUMbid8TWr3R$0NMIx z{0bg``8Qdg3%AWwrmboi3K6<-4vIGR`Umjic0T?z(`KbC^uTI8@6;6Q0#{*yPkm;_ zchUyTq4hr&;QT=vY6?FM0)cbNnExRi*JQ(Ovfy!tm=1pux_EFGvyglgK+6F@C3oVJH!+o97cq?B*dm+=(=L`Q4``U_x=?F zv=n))I6+UI%Hu!vX`o&#Fd*j3VBH36+DA_YhT-N16krD(cju`GC|1F{exP{kuw09u z?RH#;My7gK%c30*Echu;nd+r2T&?Y?U(FzztH@&=55HZrD<mVZVQnpy28Hcc47N&zF0g8xRB9J|HXfnkf{wf5vuGR;+#cm-WR7VAWG#{pQ1Jg_G)`ax;OHd)~wCCo*bOWSHOM<<5+FAq+lm z9}I{=4_n%Nh)Qa`eG*~Bm&qXAn5c#tBC_F6>~t1V5lbcqIK{=BCQnb?ASa?IbRQ6M zE{gM6C>5M;2$com!Dc8a=_kgOCpWb&bTSBHCV!E#8_L*I5vXx zaE@aHk=w+^1tP>9QhX$Ow^nh%mHp^?J3upo!P;OTJt*{TfkBPbh8u{*tej6@WTWDCB9kKL8Z#aX5D31aT2YMyF0@%3f>^Ty1x+wAH`b*iPsz zZwu#yHeV2bmZQXNrBz#nY5ax#K{j>4?OU2lp3WN17TZT&0NF6}an28M^6wEA{UiL-W{E z@E8dWXCRIG)WfG=8G2Q083fc3@a+sMM)tjtKbooSi%)9o1^y7%SZ@Jj2Y8~GP|ub( z`t7(6>sC@?9whZ?hnZH;Hgv6zK&3nPeC5ulDr9g59IC=kH`P*!&zO*fExdY;9yo&? z=bV3F%DW!;5gOcyO{u&4$aewzq=$qVn3;q0$?pcMxVWI2?2`<~nb+I_q9Bj%WD9I< z!HNZos~PU}8q@XT;t9kCiaE9Q21!O-!v>&(w+){j*@&lgm5B1)LO=i9`I=YeGK*qeNB+OW}Z1UQqGv58Sbxt7I<85A<#O z4EBpUvRfoavCY&MymwAH1<7YREXg;0eIE^~x+S#(_RW6lI>wouL;>YPzAbmr39HO*iw?6tBAHYgM zUj4EVLNMU@cbe#0rl}@UZ^4`=$Or2YYYcD3yn5*!Bc5vdk)uu4qUd#VT_&lPxJmx+ z$7dW6-5egXZ57Hnrd2K^uZ7%3(YGL~BL&unAP0qJ$v}gj%cNQ@!6ie_<)tSu>l!N? zWCzwb+)qY(po}1l=ETIg5$xPSywcjc?*FnCQ}PqaZ=q^e+u4W7AlTC*96>Xf|bUn(y zT+XLA^zdDhK1w{IOk*L1KcY)SR|JHjq_WaOmk;sHa6{mdd^F8SQ&~*%xt^CtJQwZe zhJa!(E_3Nb92_R_y=^ds*NkX4wK!_l`7 zhU)Kk6tNq8^3*;4v@Q($RaFj2shTI1%rg$1(MaK#NLpUsYvAj4oO@B9;$saq{qN0> z>6QvozmCI@>6hgVWDB`k+J)QgJr#?oAi1HdzeVW8buqVYEcS{w>4~1Q%;_7E1A{Tv z_3@HiiO%ITboip0(h=s_FdK=AWhfl

    dnSc(B9@O^@O$ku6{J_LOv9klQy;GSPdjmkXi-0z<4#KyV27uMq5T z>Fxu~z_1wRRg}B0Jj7yqKl|)`e)F%%z}?J}dm_;SO*h*eEZ(W5Y{b2aIXK|`QhIr zLBwLkQ ztxVmEo3Hmh)lY(?U=JePy#5#rl7AUc&|farhBDq1A3-<**94Th$PbmrS$;{6h!E<1 z%vgro&R4;t_^o|r(P^)ySni6rA*9{VJpM*msjBk2l)Sol5xfi#PZmXxCFlFfQvtNlEDQU zXZm}+l1h(!bG?vHnS^M&$xFcW4VD+xB zb8b`sXe`yl1p^aFO$`Hqc8DA}Hpsw@tfpQFX*qe`zM#3Ln~VS}v;nMV()_=03*z7) zsnYLQ%dqZEy`5li;N%{9=u~V7%9hfH4CHNo(1q_lQys=CbDg?q>;(cQ5rhXsaw?=LlXVy}}CN}^H;cqwU0M}=X0&%f& z7LeqadGu)>;K>!Fbbx_TViA3j)>56Xz=hoU-$5rq)=z0&yC+wVBNP*!Tz2nsQ#sqK zJ(e!;_{$2aX|@PbNM9m!^71nFyhjN9(3u2W3gD*A*bvZTAwK z;ZBaZ!><**nC!AvO&NI9{ErtVuqrdOVgQH}I2nIeB1cB`U!UQcg=yeLT*3fp6uF~Y zA_VHSlZFS7J@XqdA)jf$C+Dl@FbyBae8wg>TbSHOZbcBEzFrxje%}SB@K~RHT~3Jj z5|S(us}o~`>Qb-X@KJgHw*@rqnEx-s(y=2b+_XxpEh#ZBU{t^fBVXwYQO<#Ei=OPS zpF0)YQpbT5@aFz-ru{ zC2#w?-%*@K+jl96&lGGrEb4(C&b%yKVxFGxW}z1~t*1=VVo|2*x}MNpN8fhubqJN2 z5h^j#Ys(S$jkikHb6xvRG<%_0x`vW(57PyO&6P{fu?}qi|9|_}KTn~>fhk#boma{Y zW>7s~I}6(k1x4cJAg1YMsZnsTaw&PKbBf(dOCEwAEFl-QXECUMGvME~npl`yE<P?seICK^W0#WU9SBO*Ek<42 zlFZK|JnC#?fN&1ud(5(iYs&W#C;W&`;(L~Sa>u@Z)!%j>hm#jP%+hgI5HY1AYSFY# zwXVy7`vW64dorwoQklCW0|}m{NLC+g7+UrfSrw z8hVNGqRKFT%U+(c^+|>Qhf68f8JWkN(vyh}SIuO6mhAxFsRwqHFO+gC#|kbcc*$`0 zX!xJ{jL+frvZN2Z_v~9*)ZuK2l7{>Evr42mRjGDca1#g3ha(WYfCeW@%2UTYtbyM6oG_&B&-J<>WtdU2Qr_`@_7b-AO8o#^WY zc4)cH(KH(6kZMs3KyYoWtIM0xC5$@BmZH#<0#$Kd@0GLrZ~wwn3|*NzTjHKJMa4;b zX543A>(<2wDA3U9a%ys_l*$e`-n<%+dPq4?DKdery#-FJbM60I`>9ah($}C457dT+ z-n}ZscZ_V(qrB@OLF1#m>taj>jBQ81XdraViHq6jJKaWNz|WL_A=v+SE8J~YJX*9nZ0UtBm3WSeYIq8UijD~vku5CM;v zY@r}2d>G89vefD!tR^|pf82Vk4qR2+TH|HAc_6#bSyr^c3hM{>NoK9Yw&inVmAh#kd5s^1zw&Ts%$VhrZPgUD z>MAN=gCnCcG473n;b9=QZI#wy0epbbYyAov`x_LhpPj`bbUDS!9-NH8LI*CY#dqb= z?cSk3(xrUbK7WD0Z?^p!hL^t=QNtTUJhnkRJR$JMM! zbbv3oDA8KX97_jays2=GG`3{XwUvr0R_|2wWXB`RA7LcslNXLyN6}i0p$b+ zlA8PYqnn9M{Rl>}vG^uNhB;kk;lVA}J}IwPTXYsY1fbEIrD?gwhhBP}8NoKl<{G?_ zMxIa<6d8zB2?m-al2uHt`%Q-p)u*>X@TJrbj+Ba3wI@)?yeDbetgT>-=wZB*?j!Cu z*a3&^6VakC#K4PR8>OuDG`*98j5Y5fGYWVUZ}3oe^B`>Xtu`_~Ao>9nxs1^F_u*Rz zJWRd^Z#GTlHdMF0^aC$(FmCDr1GxEz?vIb*x1biUHXyQ#YT+0`qsyw9Uf^1WtkD>! z@TM*rD>PmP$(LP!w@X6UQTS0&_x6*iCC$BlX>s$5yWm4oZqR0tn_~86_*{|*Vldx; zSaS`-+mNR$jCQXOhM%yO;@@)EEa(#Vi4JeDT;m>k$qjFEo!^rvqKm2)$qo>(5rW!Q zg1o=P#KjJte!VJGC5}1L=#4{4Jh0wD!Q~7USzj^u_cBNq49{%dI>LvpvKy1JFm()B zEh4X4JZmSiA$tHG$s?K(Uggu*3C!8ZR9>F~!04>I zly~)(qC8%!e-=uZ4*F=9aAY2=e54O+mI$vPq?+oQwi6P)LC-}J2V)T}woaw8zK18A ztGnR1{SjM*azHAAs;#nm(l>Q&cCH%$@>mG=W<>|~8`bm+m2eg{OgnU%m9R8sKw|Y< zvX9gSr5UzyRR7#y_78i;u`i_7E<#whGP$^DG2oS2NHl!tOcb*96}&b)T6&?o><(3IQk$8-*J;b#>~Ta-8&zxjE&(5mxZ2dKJEu`A zTSC>*F-9154n^M~t{nKl#S!wSd}O@RV+Pex2Qr48KJ zD8gGJRtAj;fLspVVp$LUg#{E)S}bqiWS30*K;RyW<~lrp2Hii-x}KE(C!nNzR2y#J z2m`^2JEs`9B%zpR^uz1%y{F_>L)`7;2s+8qyGevkk<==o^R9g$bb4Weh&*`}&>)*I zQ_M)aZu(g4RWw56e4@Lgt@#9(sT9SEH{6g4c8+xjrI4|?MH6fks%*4@t)196!kkx- z?QtK{%;r(|sB%5gjZOixc-5oI5-KM2hJGpHHelNscBLN4gnVG6?dEK$BNM)}(ic7f z%B04<#WBC?tl|2&AERkL!RkPkDJ2%K*Xy*;$MDiw{VoXvSYuszVpN#*gckVM`U}(C zH2##;(XFO$`aLzyn{Df9Xw)gfb-#1!*iy?5#@H$Pb0@%>B{oALO;KK_O-K4#K#g)u zhz%!=mmUm;w6l@<09espBw|9$jjZqtK24u{BZ|9u**~Zq>@sju0FoR=I4Q&ABwe_? zsp-j(RDAf1RW&VcpQOLxWCT;i4$5kNE`j3g?W|}`&hm;~JEjI#8`a9#h3${3_Sjt( zXw?0VRap6D18cV8|EZ_euND{D@`FVgTxDVwCCNE?zR=KcBqL!wHka0=&|71~*JQe% z=`)gope_aabtsqSxc-QcEz|t&rfioP;Wj zC{P|PxE`3JfN-0`jF)|TZ0V-LedbB^gbve)cp~hQZGj3WpoAbe&Af2~gEgFEkVYR+ znLir{b6@B(({;Hx}q)AhK^Ck^RxEveLIooP>|E zSLr`^2X#YF>9N#s{|Q0@6_dBz8d=Rm62KSV4JVZT;yn$cueOf)!?84c-T>(bc*wGh zpdCS|lc%UKj1zBhGFI;WbTg(|XeRBF3Fvv+T&SNTS#aWTO`~b#_pMv85!D>@)fG){ z*imu;2=Jn)BzaANW2Uv&aF0md2`kj;p_XL!mxb7|O;fM^tMxUW1T;eX4T&$OI7O%g zIy#Ro*dVMdJsptUxMGzQIW8cU^Y&5Mg6thG!q8B+zZP0I;Sl`wBE6x(DUWX@%**n# zDwV!Vq!I4}*xx{9w)NXBA!V?4`=(aO3BlwH+#RfY+%h;{c08BVE>fn~ks+kNvy^2v z8Ti}wY+&W8Ys@aYvIsY(+uJ1x0-`8Oh#Xefx9b^<8GvLmn(n~Pt$J7E&QLF^-^{?! zV_vf)T2*VxeB%9;H@uSLGh2rTVEUu=_S362L@NVRX<>PPmdY$Rh(mbD*Yv91_9=?R zR`QD3XZ=JHWHAlN^}QC9E!j(T*yD-eDjw^*WN)9oRQt#Y+nb9MJ3<{&0%TXs7hi-2 z0_8M4fIdPPEOAyr4n==~8d&;Pf%i=2prqY^(NaLmRYpb9_(`|exuwo#mk;n1BqkG2 zsP-!_jgkkmr;7L9I^j{?pu&OkBm>UVB2eBRT{(*~aqJTmrY)dZj0Kd`8ZSnP6>q5Y zmRqvjvWfPhA?_V~Cr$)|48av$TO&1@~i`p1wm9{W$mV~T{ znuH(iwpb9P&_93T@goK|C7B3%TUFmrQ}1kL-nnp4q5wTb(iMPQmBZ5&OFnb^#Lwny zp0&95=V4+`#ri=-(z~bfP>rCWrzMvA`Im;GO`_S37-tUw>tR5MBg!rGR|&SF{LhCX zRnyz#$s>w2FZ|2wD2Hil!g`=F+~Q89^bG&@K)7GkQp?BtZcPLbyFR!m_*3;|7`WW3 zCzVH(U5+4s*crgI(~xI96XhqWc{yXo$tm9y^Q)RxdS8z*pF*^wLtMmNM|)>dcq8|l zl$k+m4N1n{$Fc>xZcOzqXr&15;0;LM<_H!qaK(2W@-B@p`@gyTlZgeSpc&?Ix2x9K zck>?&zOyCqft68`*z3vvF?cxcAz{5I?LAB;%k`z?cPyIM#a%=+`wA>00=il5XUR6c zBHYd%a>YBr^}<;#7=kZ7Xr2}4k{I2fcw{qzR@S^2bBid8*5FY^d-AOcAsf3rNJi=) zGsQu$7rZ+GQ38w8RcebROZI?iU(hyaVZDu; z$MnGE9hl2w7SzIsqKJ%^dWPwn9A1BHdE3wsEO<3}Nxj9j;JactNOx%fhugAayq3sT z5uj#Gp_CO!4LZ*m-gM9-$hJge_5O3lS2ls@qpa1RI2?;bnRUljm0}t8&(y^8Yg4}I z=+`uYKr@J)_|K{;C6LWUadv+rWU26l|5|HI+OJz(nY(-b9bPJ?E1jzF6&6(q&j_{# z9c|T~7CRnIOLz{f^;)=DjFpdAMz6r`^CR4>sa%?G8TkZSAF}nkvCaD$*S2GiW<9>& zGU>3i-Mk14$(df(LM<5>G(jO6pRBwp<`Jwi=flvHTd(t;wqeSGWlhl>5wEbDiozv` z6}p{mf~8SjSOcIm?f9)>{Mi6EK*+ycLpPLZ9J!&~3u=lnstbxgQFJ7^jxj5$2)BH`i0G*DQ4s**Lc4azy%Q8JGOJu!7K%H*Zx!V9-)~L_Dsr^AWe>1 z?cTO`^=e}mF9~CceLFY4Dp=@F)V_&T;l3C05=9OCY(U)(4;klo5>&hE6n(n4Azs0j z2qYD==@6dwfYLo4UA8A!xmk0g1j||T(bZ~ zwc!`D7umH6(@JqW%{X|tgKc1;WxdFfWK;|cxb6dgEhcZ5O44r@bkiEnvbEsLp(W0@ zF{_UhV-*lN7fje2_OB)bSodcFevaboSsAYG5?wCyQ!O7YNp5`3HGpHH5jUTKGX(z$`b?q|cBtV6R0=P$glWvG>@H z@=>q{>NI+7*NIKh(9Mr!pGHm0CK;3Vjtv|%gA(sIUIJ?=jR?f8k_5^hWY|6Xk}+DUt#mNILr|uoz%f~W{%{P>HxbQ<-tCgN<2LnyGz36B?^(UQ4e7oAVf{bQ=P!2g9a(oG9}hU`b8Jk>L5t(1d< z21$#saT>p_=Dv`Jr!>W=zCk{rf?BEkj}w?GNh4VcEUuOn26{HG14vOvZ$jSOnc2*` zWyH=Ts`uZ~my93+&YU-2oa|W$R!C~2uMQhaTZmHy4j&SPZVGQp_)wh7`eBDtZ2Lb{ zzgV69^GG)tBvln?(AWkX=uwpb#1gvbfB*-?kV$T_5?1{`NHsCBWh_%n6JI_DN|-!O z2{C_#hlm(i2gEi0Qf4kym~(c~$uru!1`yo9%TM$<)^UaGCuvwn5QnL|u&14s4>yD% z@cI7iEI-tR0)p4muZKjX)2xBpn0H3m^O4<0#&owz`8xhyAo&=KiEtgTOy_3_&&w+Z zD(vlF8RsqvyT}Dkf3Z?3Yu@am(n103Cy)nD`m|kn@+=IS=3MY2eh6*t^}f;OouvB< zAdE8v*4e<%(;{^3@rfyJq-l>cp+FUNZ7-&2+5aD1#fLcLH5Em54D&Q)i&u}dLGn|* zSh7j?k($p)DNyxwq=`Je@Um|y$r?R;Q6thyy&W}vk>qg_i@rm&ZrMdcO4S<>^neTt z1Q4|Qomj&dYH@ba&oE|dQ5I(0Xr4l3o9$qA`-L)?OCXB_sn<&Xrd?c>WdBce5(!ff_D^h0FNQ>;z{mEd>EpNKUD}du#^889tq0Obw|!7z0)~`V#KoK2$6u1hhlDX909o~11w9v z1XC~l5g5beJ|nqv94bCTzpvItr2E&r5CTk0)sX4;=^lp>Wuy0kluS7Tj7R+}8k~nY zX;cR3(Sf_3y??C}SCQQ)MGi-*A}ok>JNZHT{3 zNP8%pv(fk)6n=G`tl8PE9;Yn(wLze>83AK@Sb{CQH{B| z;tm3aG7wY-aVYam%{^}AFLSFAyw1ommHTV;@UUe)lDF{JR213#6W(y94ogW+!!L$7 zqos3=N27C8l6L2V>)ojJ@;w0Y6KM)}e}mKUwUg-`tk(44eo+ z`A;U{8LZ-;leeV9uLqqePBnwxKZ5$}yrpXy$G|!!mp<&s@l=Ac<3C`MJg8eup#YVw z!6mHgvaZ8O3d1?NS$=hA#I#1hNw8loOU{o{aA&fI`3#2jptu@}15mY0zej4C01mPj zEFy;&#&Wo_p-MX>R@(-$Y$N?4u8vMi^m45Kx}7$bbvwHn@K^wfkH1< z)dn3f1*0?|PaBi^p)-kP6mwUdov?o6M=7e7E0jbs-@mf_RZ2Lt=u>|{hD1)iyJx4X z=MNUIy?UOK2j`o!7@@$KU(Q3*{@-Tv6EySS$&8#E{Pc!uo7*CGY9MVX1bsBBCQZJ| zo9YYmsB|VDpJ|6N%D3il`j+K+Gkw0ikuzEpgQJ{gjcs_rJk%g6ErzsGE4<5$vw=~> zL0>kk{(Oqk8QMlJS{e4SA>H?CXxDzShsZP$u5b!JJTcK{ktK>r{9nqFU>BEZ8x(Ni z85B4_%h=P@RE&=(11Hut-_B(wJmc5|i7pD}T(^HeSlmi?{(a(o;)BQaUb_XY#!;uj zs^|v#opPwDRz-Cr6-Q*tw zcQMZop-!D+r>WMQyc%kE@@0kZTE_CNs!vLU6gMHfKO9b}!--H-jA8j_f7IBephB+m z`dN?|=BTBxSl{&&I%*F7G37jLYHHruZayHM{S$-Hy|INCLIh>_qSBbB+6XN6ic;@U zC->KlM)?63*xvC$C5;*-w4B^Y@QMR{B$%-RfQN0kA z2^GwPbN9p(GBTT3C7Pmn@;3_wSyI6Hgm2DKaBuxQ@Cj#!IYY%Yrde3wOXSh8SslK! zS>t;R{1kxaVXv)wY@Pc=;S)~0oUA=2U%;H#X4CKaxms!s{%qVW!%8oKJyBLG# zH<84l4hN)vwR<8(r12R-b?$l4C~`&2AI~xDWIrIH z>)?baO7hrd!VhuivBQw$(gV1@DWj8wxtkRPn*e7 zH93)rg&gTnj7n3xt~(5HEOlIYzbsxBy9pyo!JDF2Xj~|n$&j$`U9@a27#$w?O*Y23 zKH7fB!+yb3xcj;(i$yw#@`;A3Q6BFPBb!@Q)9mmM2+$98QJfj#ZA>oS8?;_l-4S~w zo>E(bJ|6v8eU8BHAz{ei?W%4IW&AA`Fxw;SH%?TkTvF=*lonz$2>c)VvgC$p`1-Wz!8IUw~5yw58wE!v`9bz zofji&zAY)1XLTjg?V4?aXNzPw&!=&>8;4`fB>?sTM&9t=Dr972W45?#=j$FG1=3eo zYZfj}Q$SNy+mrme!J&QpJ|oa)|6-H?cN@rkV>g&DuWO)iyyt`@uGxKS?XuGPg@L+- zQnfSHB|Z82>w^^8u@}FnXN$<^{ zwFg`ZmWJ6H>Q-O+4vDOl7(;*X!8}MM5Wq?L zhC0xd{z6dvq!8vRrkfK-S#J#z05bk3G3S#2)ZF!agzslVL9kr3oF?MLC##= z#JKmPk12uKYY_8xKgWc-rFwmnm_b@%hI!*E*_m;yrK!X&af34E-3Bv-aD>dUa?afT za7d(bM-G7->SnqbZ5}b)x;}Vg@#B;Ib@&HM2OOc8Fly7rH9=#TB>KQ|iiydi0nwbK=BN8BamAuqXFCMN7)JjyA)VTrYOJJ7I6uVJC|h=XGi7 zvJXdKH~MDrJy0pt82n#b0e_<|_`RaFsZm?{^h28K&HkTV2bxwTGCD9(NRJiwc8J`t zE94uilAGKBp!}+{sxG|hjFmhIx&SgGXDa1i;0n#trj-z*R3J!_6~AsNEtr~mH>t}_ zGpFZHZG1h#Y(+=;uB{hz!qR|4{>vXz%oKsP;^Ap5`nBW*(ZB#8tv8~E7XwfC65xSo zPZ=);%~AU`1Sn|8eNOZ?HIe*w^;X9$|NJ$2vqarbX%^j0#D5bne>VA>i!?u=V;+hi zk4!R5roKo2C{;}v#ari?dWw&&6B`Uhyk;p7N+x!#9XP7%fRY9-e0%Safxy%G9Hb3V zhM@Z2n1@aauwLU29q$ib;`Oqd_k7ufJYR)&Q85EOMzVcNqnI!({&NSol5I+O(%DvgCmG z+_C4<(I7{DaDDyx`$*@)G48NXs2>+`JGFt17O{0W45dK;b0HfrrOT#!dl=~$K7_en z0gjOf@Wm3kfs=u~5nd$>c#KLgj7KiUVg}(jqRmD=5X)a#Q;#z`kUkx1mQ#>IIYx2mop&n? zGput0xgAKdo^z*tM}Rw!K_f0IHk73hG;{k2=u%kf2nL(qmB$VKQ4%i1NL6LUgua_T zxf&|)SjkHU>ozQZ#HI|xVecAyNaLnY8Cq$zbB7N`t(?DZUavqA1(2iUPlP5|TnbIO zTBul`sgcC-q^{pX{}30l=O3WEn& z9msBi1gS||D7;LTsK~cMjLzrBHYIR$kZhVa0*I>5K>>pl$plM4CnL;UY}96*m`jB1 z;mbK4=Eb9k^d^8D@|CWEG9IbL3ORsa-rOcod_5&8?~j>#7eSOV?b4qxUe-n+e5~)i zb49(ixG?b*Xipv^6BtjI3ngKj53{wdC!ahRvkI6dmb58ssqVhO#y-Xy;8M@)6!c~g ziD<~(`1i(VY%ZQVJ{%!zcASdf9`Uhq@B&3Nbq{!*U~xx@b|CGei_JNg-4*zSpOdxW zQ3vs!{M#V?2?dch#;fIw!pqiV2BAAkgfMPI^``ecmX8%lj3*D=d2}xxaR$?Le5~m? zQ4~a1vxx?WW>6#wtY4KrDk>Bp14V3d;#PK{#-3JLU$sEh{w6K4cRsl*mUE3WZLbxS z**NIc$(03@rqOWpBn`b2p;CQI{B@8qbN^I_;;=w)8Aa#+)e051!_<=n`MSR8put$C zz{_k*Env{8jT8=B%TdC=sDA6ahm7e55y-dz!LV|VyOZKZL}WJr&TC<^IZ=DQ)(acJ zwfb902mk<|e*#ZnnTF5ls%^pe=`VdYuscK$oqmoir=v&Wdf5bbr{k}(1=CatSuy!q zN!MMJWz)L2U&|)KUE6#wM6XjANDq;D^S^TgQrZ9YKT&9pqQs5dqV0aDXL<$?&lmz| z`SU7q^`$sAKe4mkJ67`7PBme^p&b)C2bjK-OZ$B%Nd>Sxr;hB|7Kv`Wl`DXBV=gd? zjaTcq5$2j$+;{fKAS#OTCxNJFT`y9@$97^Uhrgh#KPY0E8fWNCC?4}3C!*uX51K(T z1G^4J_Vlhuiijf(q)>dc}<+`+`ZiPoc@nSuQ@L8(5~P8vrxwK^gN!)~hmU?8;jTK%rW2_9#|abrG^fcU~Ht$H+VNYtP?{+i~GcjhfkX$Gkpan|;5#Xu?LB5xjb!v(B2^ zi$GA=fm88Iv+VDLJ|QhPl0OrroUaQ?-Om}lXD5!*DjEC1&9#=#Bu3yAiothsTkbkH zBLG-jii9Z)1rf7QU}Qo}tNr5duHvZy23@O22nF1Q{zTtEM*aN5ewKcorMX)J$@2-+Nkp zQ3>g0tSo@i<5(_%!~(R*Cof2n$0?Y$v2vf3DBG2eIJQMu3OdS)mi8N7c9h$2OAS zLu8}~;?SwcApQStzI|Xj1?Z|4#<$KL9I{4lp#%K|qc1~L+CCiMQ}T7gbAKclQ_qgQ z1UnC*-#)BslM>z`&^_znS?vAl)>$j9hd*#u%ERwz5`>zH&;KV9YM(g7(LL*Nh8QYJ z05q(|RT4_0orvWzk7m(T_qRYEa{R;np1X0V^WqMf>cRB;0}rum_V;uH^l70ha{Q<3Eb7&U7EmXrK~ut zyHa$tKZ&o8b%mf=YN{n9P}R2D$rBi)S+_#Z`V)F;`i*!9h~5BYdI|iZJH*2@3{j{> z6ZJS7vh=@gGMh@pTD(NrU#p&c%eH^Xe$(Z~)gt8L_s zch_EGj<=g=N2qIDhR;N8OPwvaO~nqhNm6Cj4c?(;^^#YpZjvpPU!tMpGWFa1TEykh z)FMFd;SW)l7&Z=WBz$Z<;2_wGFDSK#wJB-Q&K_NU^3flAEuQ{Ut3CV_q8mZV;~y9j zL7k|XEJx5&vAf+U;ppWiVZrd9if5KpV$L^?vX1$Zt^EGTZ(+v6DtbG7^0)|Y=a-Bmo&S}_F_04B zpk0xHZ1Jyu=@fI*K8bbz>|P@)Ou_R#s{sKvoZn+Cg7ZNJ`asBB-^X*mQp8&HPIZ!P z6qfj3k6q9uK=(HwykA|AOH%rsf{%om>1Se1i=&G?o3B%T?8iI@plk88cFBKl`JUiK ze4FoOjNwQpc+V!svZsI76WPAF&`V&}#xdP%18IlaCUsIyg__q(ynO}tZQ}t#7Q&o? zvES(%efjU^fnaP1!$Mg?jGcfVBZ!fu@xA0D)AZRAO8V93<&!Fq5lvFWXA3N~`G)BW z2+NeL?I#7mB#&QfDJF=48iN9<~tNRF?BLnZoXTG z?n2y$SlC*JRsGAX^o0%{W}KB^f4SHkpn`KobOLpoxYZ3=*X4W(3{rKH<@#QdsM3?X zkXevC0lJ&)CbP5*9PsqF5GC=KizQU_!@p+Q%}i?f|4k*Ane&qCSB@iJxi*r$&iL29 z7Gn;HOJ`NiNMq&vzfSRPlXrVo{X8t>6jq?{r0kR*n-{Whwth)NI8MuSh_>^Zm zRuh(=sNdWzs-;rhjz=kezaMACEp(wku71g=>uw`SMr}>n#T|{x%>CeHYF{A6*R4CW z>lJ6dWCgoz&|rs2wGg6P@+^fadYNnW=*`F3=6HHKmLX|g#U|-qlDn4nAeH%HpaU_{ zT7imXGX3OPyYZDjV3FP^FVKR_MajhoGFJppHWM&y#*#9mM;u;$_AL-L_*?dsF?mB1 z5IkH@De87%Quq--t|Mzim0i)Z1_~JBB1oLynb2W^1?Bhla6DFpg4O$XQ_DeXAcaFN z+RrmYVKA1IGkX+GKmhW{0o&Ei2JPUF13XjM9z(dLNJ=Z6%}7SJoXuc>#=Kw(p9-i< z4;O7qGv2=JNi4$PAws2&d8iV@r92^!PP9IAJHbtH!dU>FFUMLDx_kE?;%eBKp+e5i z8N_xwOIQv-Kq-+M#zn5MHwq>4YQ3pOZ8d7=!(RhleLqLXYoEH>6;ajH@cO%it~1}2 zm97dPgXRj0EILsFa1vWDC*%#LDNQl0PPZ;W9n%Ct9J8LN!JG3Rn{2Alnl?Y|N% zbFm2qmcBNHNgJbu`GH_+h-s9UbbBQD@$~ia_qTk|@0@?~q8nad0Yx$d$?C%*2BkSV z)z~>~0MP=SG6@Y2V;E~_-P4@O_&+n@8k+ooUeu7%?zCKN6`jr@TIF6OPfd-a1_$2w zObYQuJZ`D!tTfL;TtF42ck1uefJ>NTf0Uo3QXn^|^H#s%zuLVmN#Si_e{XwO^tMDL zoj6Iw!@BR!-L%AD9&=?0x{Xj+0{>A}?_M>W{VV+A*!d$INSG3rJaa&r`T0%kQekE& zpc-U5(}4ql`N)KO{v#^DR3SH+xR-c0Em7zkpoiR()OwqZR69EJN-1W6Nf*P@jL9@uOO>w+G7OChM)a4`!=td&)^~6~ zsO@g;qhs;i=!w$WPpxDP3sTgfFeqz(aPms)c4-t~)cckq-xctb74%5ve#78$5a$CfQ?6}TrB zHD{!wFXZppqzebEuk?X~O~BwcaKGNrzWaDw60C`t3VwBe=9V(3!=$$+XrQDv*%>ws zbCJKXcQ6Y&ls7iQtb$#qaIMrkqqX#lLq0G(;oZs6uZFQnw=gMfy^jTepA6OA zUX+ahY>m3;!@+M`rzyu1IkuO1Nxwc99ykR>tM8vE5(a$k>aq2&zfY&juBwGR3o77N znbAhQZX4a&Qc6HupSuI2t@aSsi5zELjgjxW3hU|pV6uL1cyt)C!v6<`LXC7t^qr6s6-r}pa~3(wfy;A?BEB=;i{E8LIF%Z z|E-xBPtk!^l#G5#-tggqY?QKI9W8u{juc-_du+s_uTtc>-f?6&{AdB%5Lz*`tEpih zrPdqbSoazBM8L=s<}vo3gffpG&k)=*`P($b|8QN6o>pVtqpOAT8*uNBV$5uBX?(P= z|EviR#g+P9*)U$B`Wb6Lt5U#r`TR~XS;Zov48h&4VUylnuOaW{EW%O5w3c%5QNT?3 zHTNd#FCGVC>ylQ-U;67sCa_tFQuava-HmMF{Rjw-z&2eV(4Qm**P^3x%Q6Gb+Fym8 zwnUNBK1?XPpHWs?%EW9e;T<^C`HF2F)#S5zPW$?EQ*hl*Ix!@~$> z#@2&3Ecz{$5e(5%EZSV{i`2?|1mPGRtF{ZvBHtTI2|*s&ek^jrILByrEx*JFSH|I^ z9tRc9vKeXJ6XY%MYaSi`hm{~`HODb&hOXRI&RcTH6`v`+DVHpKv~B5a>zBOXd&qqfGVI4 zkc}rZtzdS8S8JoZDv?5s^yY$_S}K;hT>V^8Yo4HQN;xyHSzi+zA1?PXj-I1}>tV3g zM=D`;tqT|I_sk6)IK{MmywM4*HK1I`G9%%QpmG{XQ9`h=2#C-wdF#1)hT|0Tsq?m-M7eeb$Htit zbu>gxq9WzU@?&V_dSHDpa7eI!rOfSc9p{-l=vkaoG}vMW%2M zw;yjT5=Q%5|ERd@d}QSJ5e~Y&+Y~m7wZ}QZA)C`x<*#<9qp zi>X8<>^l;`FJ`5e_{2GutM(0Q09w`pc|L9AV{$HdP>eLDmlbC~!;p{g+*DxZ+zBVf zL5TPR8h#~Lpwmvp^ewo^f)uReESn82Qjv|cZ?A)%-ot0KV8*H(lQVADv%yFTL2L@V z0()y1LyY~Bs*}J{Xh1@p#HiOyzwb&B+EfzMLCfs=NWo6m`_NGGE=7!ggq{GVJ(lT2EGo!@)$?2F zQeKux6S`ddm?$3lM}zxP$~Vt2+CqbZj7YJ5FFb{T3-2wSbD^$S!)J&hmxdbc?#F;< zhL#0Ho0*nV2_f|tTdTKLSToIaR2ZZ^asg-fIq!OgP-PoC|DI{)R*H9_YTpZu+F?96 z$*|Pqjg8S&+T|LCL}H()v=s7oWV<1?*WiYvX3VL3x}vx@t2za~>17_HTEGqvtOz zmLpd%qhJ>kGO1wR>*(dAcjbTA?Yps6fBHs(hT-fR(qR_yKc9G^6zHA=j3(#~u6O#w z6Cns_p)yCxXhHWIfB$}BphPx-aCxHlTs$cUHP!AI2-!x17N z{!%Nl!U{$z zOaX!>@9GZu%0UANIlWvYW&|vFPKL3QDcna&U0sz%?|42fgetm8+Pc* z#_e@=GKZNm;ZFCqAw-wt~sPF_~ekIh&fEFVe9YZhx6v2#3isR zyMAM!@WIfvYl|5=sw|4`lNXXPWncr^?y!7+iP9AmA2QYYYGa@$lJYL=6bQ5tBSDgfxsw>Q3pQKv=v}yL9|}WY#t(`HS=N5BR?q6 z;wCci^W%1KrZ){`X)Stl_?O%g+|IPKy#+Ny9If-QjBNY5gk+!-I!zV8S`xhu;`ljF z{?4T6KM7S9F`A;mb6a;6^~oL4vtz1(p3K&y4Rx&qGe!Z%0$C2q$ei1<#V;=n3cC0# z=15YNgPyRB#$3ffvGCpV3GlM?)&OUIY7cGTCC0U73h9&rpvc}JaW@YaZ*Rvv1z{=- zlk6?!pOT&v-U%PeTZd+C?WyVcA;X}0df#8BE)@D2FUcx3kpRASE}2W{(7I#}9Tq&x zXCCzWG&xm)Hva;ze!p1(Bwtk}+U%soM#wyx5J(ix+bde@vA-pifuk=hMQy-G!Ndvq zA9Q>Yh4j}y@|Q+qKL!)PO|#!V2MWibTLdvi8vM>l$PE6&FZa)lu4BqVG0aGf_5Stn zwqC>a#n8-!*sBn?;TzP5&p0g$d?u*U&^jj)?i%y^Z(NhN*nFp-u=DJ#en6l$_ri?< z5xy3jEq37U&}17vUradXG)>M{)-nf1xk#4~^67bz+F=3TI z7$%M<#MtS-=gQ1-+LRb4(7LV=1EYhHo_3>kd|l9_zyaV zK6Xfh7CTDjYbOnWp@eTohH>~Ti2T9v^FoIXyJJFhK|DDJSwMR&^4`stxVf{%?2T;t z7XZ;PTIWP(Wl>ayA__t{M$6QSbQ9insHN2D{%T?^lb44HFD8LfR($OWc2HJvM zhMjRW&8u<|&tG85{D;XzKG+L%xxh*#<39rE!~1srHx7^D0jOO&Uh{a!IVUKj=o(cz zvT;CBki9_QG^0`89(Et)0xDjn&nE=G9OIaR+Q30fDF-alOS_BJLF%K_EbPnk2_EgU zu+@maqv(U`+Gjx9q(!W;(;e2eljC1<+DP_eA9WvM`^nnv<=&DA$UV9-<|LDa&7$ZH z8AYE0%Q|smjd)&O$hfnJW9tJ>qZc7577P>i$9l33|GpbSlyOmPpu8HhOd^SD$IOvUnT^iB!%g2WmTaMGCgLfOncjHmCw6RKp3m8q^8%Kk|3bD&k~-1#^y z0`!6oNL+g*UK&UmdYv*p#P)H(@rgTH?`Sgqbd7AeEL%a?f00_r%EhWbl`#nD?LGzr zmP-0IP)YB~={taB=re1%Wgk$|5ubP4cx!# z>WvoZJxDG?A*N?&qjO0r5hsR{pM&PB{vB)OqmY;ldL%Ya04e(uI~A#Oq7n7V$A#(u zU)M@}kgiN|q=8VRalLB=pcZ-lrsyU1X zTAE#Q#Kd86^-T%pj)VaNOq$?m_7t4;P`XWU7`nkh`k`Y0gp^eJ6zmiX#bM9Y$?JgI zPh;;NJS4`(f4(d%z@ZOpz&|OMIew&RGaGQcyUQwqv6BI}js9SOV7Wf2XbIFU*Ql;j z9$FaR<BeL!dQCR{LqSO**Mx~*=IV45T5FZ!vBPJx6wOT5)BSNzhk!U?# z(ZbhD1?4d6hmcKqdvK;&(w8lr!>)2P6ZuHR$jA|6oSb{rWUIRF`IJ^S$>9La+lVOZ zsF5cW?79Rk3*UO1wDwds2Q(6p)=%BYDzQ)YZ*MDBr15KfBiNQ+9zKQGUl2T1w!G2j zmWISXqE6D-dE`#Z0Z_|58;%HXtm}$k&R>WMjwj(4Jul?@0h>6?)%l$LUAIX5LQCm~6L*it@ z_0z1&ph1{D1e`q5;!WP}&6L3rnGT&q704$> zp_^wn5!$2B{40jm=TgnURcn#5+2j%>SM=>Wol`X(8h8z? zKcAP+1=W2<=_(A${!1^^Ap`HwUFJpI?abAxZS66 zxcqUWI@Jyv%&JZ=!_&EU6)goVzBg)EqO;&NU5t&?{!;w^Wu}N4m<2X^!;X^JELU9o zpBi~7asca4WjV!7XtV5A0~a4)iBJEq=~^I?3S}U-m4{}$0?ijL=YBfIPY&{+35~IB zo?9RBDbD6huBmo4uC=iCriYy5VE&H_C~yDidn%2s8K^p{Q2tQM^VBAe@$A&n5YC>! zm)Q8A)M?xyM;4kLZ)LOGn-Pzln*!g`7PjN`JMo+h`VlA}sjIb6-F2>?Hweoj9F~?$ zcS;$9(Kc6Y6{CehjwuVa^0|&Y)xILs1RaP@O#d;q3YTNl^~5q;7DgwExK!u_8k}jl z0arZKDdm(q?4!(yOM1>mLel)eLtKH&Hd)GvL( z+BclWYT?FQ0dFtpu}#S#LFw5VO?nzNgE-QUrR0k91#r`xp=w-Mv?T5d9rm`V^m)zM zg`}+R?__SOS;jvrvML2D7?9s?Gr;xYC)iTxM|Ph5@?ewjDf?;tx(Wd(wi1co0-Hmh zRM$S|UaOPeVL-C-?ThA(zKT|^Z;aaT%cIYi`Y)@VRAB#KNsDthNcWobicLwnd4l)G zD}V}*h7i&Ag7h%g$sDO`?$Y)xh_-0atdb~$EAT>mhdsGLvMey#mRu<8`UYAgrnz5JbFvKlU33Jc&vl=wH?{(ui z`4>1gW|~19m%0Fzcjd>>;OXf{qD=Vug6gD4sIhS6vmh+n+z`SWbsk1UMl`1u(bn0% z9Gd58U$VXf&Ildma~*rQ%UEHy<_}xz5UtDjzQmMH8XL5oaGpQK1`~EPdT<@fkK$PL zzBgP&GQOAG&vN@>U5j(CeQc^A?_*M*Oy2R`I5C+V)lZl8te7(g69$VJA|8x?ubU@e z`PF#&vNFn_dx_fis0pA)yMlX0|b zmixd4jeVZ`T2{HK-7?MhQO~Xdn-VbN>9cs2PI}nnJRaUYO?_}LJRbx$@Psr*P{-)i z3^8Y}P}V8{oXQ7NHxQ$@o$s&7I6FcOIPUW;X*+$4om0|FNlieGUI)vSr{l7osBjhS zHuPWQoF5vR0y?ksQyhShh$4b%n~_%K{}Y|fgnt07zM4q#c*}nq3l;^T>|Tg27)m>L z=OG}+2(d}h)6s5(5dfK2b_jg9Kbc!(>9ZOjyIH%wG8T>>#8i=>o(af=?ZX1tOFD&( zLq+`sXXQl?txKusqGt(;AG|)RpqMm6-F?eab!Guj>e7o6#Wf3mtCreOa+* zoX3ER>Bna{A-|{xskw_xX7;j|$_h$i-EzTVq zO9=I^yyl+!2|>UBgj8rH427B5ay4OHc2dC)WDQirsD_&wv~cB?sVVc={U`NH8H1;X zD&hVuJk(>_+i@tW!(ZFxDL5s@?s>chnLuUrWDd*JkHt6#>QbX$=n6&csV1?OyClMk zsru!zcr&{TBt-R>(Is?QWkL9od;{*pKTpwx;m%8sHxHW-A}CRGJc96yq=vCYU4$Mb z$YWYn$I0v#o>2-fU(^f8XaCmR;l-Sc2g15aS4izj=9tZU&9uT9gOV@3`_gQF*A^_h z#n_S|!zUw9wiKzlR>$?f@diaFSpK`7Dz90_p8k1|ZF`GgLVyR!WymZsN`mbC;0fmm z2`PA29D|J&US`ioMd#!k=hETClmD@k*0zU34)X(N&W0ufSIbT2o$V*x=SJ0>Tg4`l z=tgdj`?r}#N36$vMcTE#UNM5JXBHXrM>p%g=h z<9P$O9Oacp-|%f)6(FfkRouL>%)}Vx;XnJP!&($H23LFZo=Y95Q0On#qCg{TyFZ^~ zndZ!AN)2@>K|cdV=QbO^u^F_&^uL9f0Q7z3cB%cKwS%W$H&@ZQhKx1&E-(OK``uEU zP)zzup5eM(c1yH)9(v;Vg&ugQzqFzuTi%6MDFOfNL|X9-lf9{+rbKa35_1R_0e?I6 zZx{t{?ZfIZ!2l;8(>SH4!WQ!Cp?S+1T6Dh*D4}bTyuo^fV z8>uLwf?_}4py-dzdDh*Tk+wjeMYSQx2}7qu=-Xck9IrsPT8xh%_^Ybi9E7WBUjqlaYQBMe>cuC*Ex->q`Jrr`Tc%ckjl3f{mv?S zjNBtJCKJyp`6 z*PG$s3`H(k8gsd*2=b~7n>o@W$nq9QJ1Ig&*s)WE-R2i|JJ$bp@;avt7$+~PeRd@7 zRP3`wQc`G{_7K^G1zJEMUz9ioFLff48=J=JW5_A@LgY=B<5-HkDw9Rzp=PsF?gW{S zoi@m0_~^$hlHtW{EEt;r@XqWNQtBjSOfsqSt6y)>Jj80K80gp^K1>FzJH3=7ysJdAmdO)*jSQWxLw%cwP1MC!})fc8wX7@zl%VIL>) z{3W1?4QN!cGL8+GQ_gbC9Ri!oF6{&k`jP(E7z4!K%!ojV(}#ER%|c{Y@xuOqJ1a!F zENYZ#^7JHa1{i`p`LXO0VryTvv4wU-n545tU#irx40JDF?bD$9hkPTkjgu$#{hoeM zTT(SQ@kLL2z!6w>Pi#~ANeU>>Z1CCjoF{QLhi85$3_s>e5gA;@aZ_zUMf#%~SWB8S zmuB6;Yeqi*mZ&z)V)o|sdH-skLG!0TJb=O<<0W=uls)-~75ek&P5;>19rz{-(yK7W zXo_{;0vwRXq=H8&J-?Ib0+A^n{+a2(tNV1eO-39&*CC7dXFrjKaqNW_ROjI;%ab%T zFh$VG=cl#MmVTp2$3FqO&P3*EPZKOP8JP~w%UEvm$VhW2yB+nhLMic6SpdKoHrmmA z30L4XC9`HuE>fnt*FiTpkQ*9p(d}pe_qYOQZ}!@Fl8EE)*{e^s!-cCgi-V}%k@D5R8oeFQ{M`q+>I#qe-{_iDO$( zP4-9gMSQZ3NaFDOwvWX;7Z%OF6!okChiX?&?;{Af)v)A;Vb^h?ER=>!giVnh$UHK1 z@pC_+drn>x<}_5JC#`gR*eYXrG~Q)oJx2+4G7HNp8e7^dYZetAF$pm~&9=gWuf_e| zwg!EsRs5XFW(Z2-!6^t}-~Yx-4LZp`R=A27-(3@wQL5GwTgtLOt62dz90`Rd_lA;0 zxu;C~G}DRnmB;p=al9!z+L64*nJkBRtUl_VNWc37VR}`8TZi62FfE>Zxv@nGIug;YoTCzkxHJc|~p3VGF|( zEEuK+td)8QJ{fzRjgLvc9mS?ztY*AvT3Y@Fj*>7WLW1AcaRgfmO1`z!d$uS1u>t(i4^G$wbS=WkXYb)at*kcFmq;6zS|-) zbcH}*mbWJB?M$5^9@Rmo!2ap0-&#SXx4UL%sOAAz8K~y(G6iVbeF)S;@PUGL`E)fN zPwbXZ(=Pp?d+x#_+`8VbYv}CSOjvvFLK6K!E$(Thj3T5tPI2ikKQM+WNe8hFzl69{ zpLXWRFgv7RS=ydK&P{eav-=g%={F53MasTm%^juc`|k)li}7ewx~sRrj!S3wJo5E) z&pN^Mc`?)A*h)^MYVw34Hoz2o;eoqJx)b z$H<)wH?7}a_>Z{aoX+23Gy)Ys5#d?ga{C#4rhb;P-h)W(CSF5kG;MCEjwN96ZlZxWWK3qlPRB`zQ?wu)A>NU6~ z%I0$B%JA1+SnMwr11&57X9$1Sg(Cz5oj&yh*2^d*cCj0_r{S&-OTR7rd0MAhygb-1 z-I~QuQ3H(e%+Z9?0ZFZ!++LI=dV5!f>3YVL1GWMqiB)OZ4A*{aM^L`eK@mPmNw8Lv z6izt3JNEa%(B(ijRuU)plmlO)aXk05Mq^_|LOFeF#fR1hZjPZ_EaUZ@8rzvFdWny( zG8$rTg^y6)ga6tVR?P*roFxoM-GhMo|R<+6*MMo#u9nC$B| zRyNnxCxwRkI|#iNI_54M#+w++D;?zJ2k$lw03`@J%|B7;TW}SIu|RgNtch$5*!rmi z6swQ%gBe_OJqR@#cK$ArHzwOb#HKsDaTY{&UGTX+{t zcC`RL5&?h8wLp(!{Q}(k*bW}8b1M>|N_ zRr1ay`MDl{ruxObZUigN(>S}qLBfGmc(p>FRf!tjWLLWh zWT^}ee1S|;k(rHjEwsGNALg?TR#{)h(F|iR$q^uReKLs{H?xQHKEg1%sRcKfwbUrL ztF$p*Teywab`1!gIao7xRyc*)=_A$140X2TUZuBzh*n-(N`YKV?Wd+sksMtM=A)(a zBXrq+Urvt|7W>cE-pR;2X&BW4Gb?nTr~e;(9p3pUz@~=4zhTW$)Em2={N~S!H9ly7 zH48iXJ31tOdOWKF+jJYnGQzViwAYG;t*~|OB;0DP)Yr;1kH^{&AG-JwX8Rt&^tiQ2 zNS&x25>ej82<`GgL#Y&p%Z7N{(6_}*2gn?iVm1tQ4GaSiDy+@qAdQBzKCYS8sYNvV*pn59&~T%EtScV(->YbJ%W{k5P4O= z+ms9>1bZ$~jBcL0bdZlw(^COjg7UQa)+#Wz&{j(-upA2ku7#iu$*QS%{BB*E{obLT z$SB_gjo^rX9?ijns9A=O%283^hBclbhD+zdy+a^#)8}4BCdv*(w()qkWM?vbs&s9C z+@nLbA3VMI<73BX0M#^c+|%ZnHxT-tT zhira~Vwe8@3&a3XG{F$Q4v=45#rAvidRpJ-z|EfAdbDq-uxAaH8qRAyt)|2)g?Lz>=wjMwcN=ubg%0mDUlU<(dGU=%7q%%H3B07n1CydOYcIpioluiaR)HouRw z%~l#$2vjE_XY<>ZHFTmIuurY7j)wlv z6eHpuRmnygb6pYN#Tql!b4Wsm!s*LQ6Vjkn*mM)+x^n>T=tRnp?2<}$`mbPuZ)th8 zcZ7Nm0v+8s-?!>Gce6#<*<_6p8172Tk?D%c%Jox zAt!lcErCaCZM~+uH>lc`D3k3DY~nW+J~4S%cBzMHp#yb2ohTQ8agwM0ZEMPO{)gqQ z(J}1s@K0aGDuWKpn??<6 ztC#0?854{NAv`u@Xf@}JpcsO6bi7i@tYDc5-zZEd>iHDnZ4lorqY<704f*}8gm)Xu zzu(61ovp?u=5R)(t}`~O+U|&k`Jh8JS0h-5mq+=p{^Fb0kR7z3ZXdMDM)oWYiN6U| zt-1Dgq5WV0^aE%pq8#mWV%SzHmi=^GP8(f1HWyTce5>u$*U$~a_AC75e?9~@U6F)f z>;nE|zFX{zmR$uJU9!uoT*XaL_2xJ(ARXr?fn*a;K&S^5Alb;zvrIT2msSVn)W$cio=CY|ALYzm8x*U$eDX5Cl+b!>8~b2q zT|Djm5_!8ZD6+%%=vLF>>5jq|La=mH&0yS$+)SkKk*2pPZwn-Uqsu}zd(iZ&48PD( zS0;Bigvt1?{Vbel7w)GO=H4_5+@VXLT>MCxl9eV*TvJ@+29@q+ywEGI+!t$u0CpI4 zCW?j*F`68@0Iwx-X&CIAQBi_IsaD|V<^AGUZ-s#yhrLaLZ38vwUhJ1MIbNT3HIqywRt%rjnpl~D}Y0TwydTf~@;++|WBim1Ng@`$!)OB4x zF@NSlr_fdq%Q4*=?psFK9vtF8H+& z5y2gw)T+R2&+JdC&;*_xc%amVnuQ&R5fyLeS6Nm^MlCd7RJPcBo2mMHkLnpr#TJr@uVPT6hAXY<0Jjeo+0{_zwSi7L@Pwc$iHHKQiXel3 z{nc=uyYj%E75!8{-01ArHFTBxE1V{lJbROjL?U24N-tImn13>cKaxXFr)RYKrEf5< zsCS9P18lp@xd3C%dl*sW?*8Hi=EBES%$-esS7m!w$wK}^dG8;5zlJ|8PAigQ*5R6v z(n?z;A(0}|)k(!X0srIMbGijNOzR6TtD=IKt?X-wfF zn6cJ$SjfS&Wl)Qtd~_6Yk;E%kxwvv@+m)W8j*Dq%V6+nSPi|KMcycPgoB7L;z-n@t zgl2}Sm$ZqRv~C=Ndq!?G zPhJV|x!s>4gn8t7wSV?29E#RP-Hoipd6%aJ+T6pyAIU^R8;)3k8|54z&bG>CE>hmo z@C~%Nu17%WQ%w8^hiE8+K-pT{WrKbac@`XRScy&8#*4TMV<<|ddxr1oM=bMpyZzSUD#4pS7GU^3y~}&D;QX zmc-5u9L_*|`2;?~Wk<9Cj;v~R)(2hgx#BAXLO~Wd7-w<&BJyxBI7Zf7u<`ziU~UVp z(5>d(f~^!>tuscX*=6?_}=n#uH>vNo&g=l>8$5R^Z%L%S zaliEwvsg-9&L@759ofAFsTg^8;vhX+r8}dj&d_s5N*6Kewy){nh)Y=bvdG!|1cRS= z_%qfT#`wM%UwrK*)jC`{#Va$)s)*bsGyxu15t#A*E{mgfB=?EB!L39F+Lz2Vs**KK zr$Wv&Fw4Cg!%j4`lUvym-4$XvHo!G>oAt#B7=UyMs;Vb~qK(vuQ$7H?a1Bb;z7zhE z5e$h6WERq-G(w(~+dGS?pACLb1OfC@A&0;);v$ReC&3-6w*xa~bqwhOoyMhYo8+SI zS#{w#-}Dg4Ih?QlvMN~<73w98a@f?ZRR=BGhl{DIE>2fE#@@Bo5yQ|0Sh z`z+AGC}4K3Gqe@pjyTV2A-{*8ZeK!ld~kp8Q4r2QR6Mo5FctFdcmdd1HD5{h7=Gz^ zK~~6+3vvLr@UHmA7q^eQy{)82#0qm-94#IX3buL36*~e(a<;Ar2{q^C@gniSs3=fC zPcC`7t8PemiF=P=*DgYJ(`3LySg9Il?JOT@%W4Kfo)cjGDNU_IY|HWkDB`~+*8i%> z{a3f;=Pb#m7k`*pr+|zB)#5h>SO)zqK{Fst^}=?5gE7iK!`w9zt+@2!3!y-PiUiQx z{qd)6D0^$&r0ykx?DGF-fpAk|0DB!z)KQP+v%zb}MIEvQv*6iz??)Rh{G-PmdK+BO z%B+f3OnulE#*%e6&V_J?^zD{Ehug|0KM@0Kz&KhUAFa=UyopNcWvS{{NOq2-zuMGE z3!2Tz_VL9ofghNr@w4Y4<$pBcYnN{$-XuY??q>B_)z9>JEPnFHE&7q4pl;mc)fzl< z*_71=@uK;Xn>`R5WoPAAY!K5Y>Q-%yu!YK-i}dJt8;Y9#or>uxVB2*!Czua@Fj8Fb z0_a?BHIpjYpR#h%K<4JDH?mX-^7G~HsxID+@O+$;pC#b>ISYU-svID;1VelW9K^!_ z0Nf)e+T4sY*pY%vZpVdK4SAZ#UYyZE1;zk83m)?4W;Izu0hBLf&jck|G?ru>Rn$1M z06?IpquJ_PWFl_s^`&i$)F66;WW+CmNGN`d!4+f?Evyx=$YP<4o6nS_zQIMVz7#)s z8ghOH zHBO26FXn$;6}ne2;Hwp5>{v@g=+SBh?I>y%Q#dvS6+R%N zZ|=@@(J`DO#474>RrJIOXNQ$RAx4T_04I?*t0_ZCmM#i2&5~*lQWW8qfWr#h+GQ&& zr0!{YHAGEq`5}7z=7iL053Aqe3Z$Zx!D>*K+yJ6=k`Dp|a(@6Oc3Iv*G! zGdNUCJkv5}wd;Nm*N6gT=i;vsfpX6#YQ&EmW46Hed#HkF4?};YB1XFm4S!72{lZ6R z49mM#40C9NM*l4Sa-<}0=}{;E20uiNh*f21KS?HG?o^iI^lSc zpkg5@P9L(!9cvVovx8owi43=j7ZCdO5T{lN61^NAyE!}5ynoO_cwHwhCwYOA zNK2?CfwT}&inFpX9qlz0p{u1I0&n@vcyAfIMj6TJYh|Ijsnv%7anLEcxu+mPQ~SBm z@>Y;KL(tII#J+h}*e}^Zv-82jTnz;7N)uc;<&~0cpUZUei|VdmR}E{F6(o0 zhDsL61E1|4-y%g|ojJ zj-CSkzy*5cp$PvH1GMDX5jvk&SL|#lxR=%;C;WMf4TE5+3fo5sG-15*E=PfOAN!N5 zFfCY{XDCh=aZR?v<4tROX!mvi%I6urU9&m1br1M>ic%=kN{CczL;hDfsK7xA7j0vM ztuZra$oc$(_o|R|4k83hwhGa-q5zyQluTb+DsN$mn7~TEK=FCf{U@%J5RdBZ3+Cf* zNkX82gaLo2+^H*J!wpk0JLKKv`Awe{7JdK&*p}nT&!gA_y&y0eeciY-V0oX97T}{n=8D3 zwpt0tQ9ut!l($>3jO(Aq?@6{|D!9J}<74D5*K$z%V>1sfpPn*Xz;r0~GZ>D-f+qT# zEQD^&M3Mvjj`@GufEKR_0^2*e(lM8?O(0r*68huzNIm8k7;E`!c>k2XAx&VigXtLO zs|v&wt56B-laqfufB;X*`L11f6vlns(O^yD0wh>gK$T{p*}J+(kw_U6^=>jA$0ea~ z51xqx^ZwV$PQ;$mw}w17P$(k~N8zO_Pd`Oz942lA0{u=%VRtthDYmaFX$zihV2a~h z(93z%ne~YI38k&a@ujWEu`jnZCtow|aPq@sc4{swz@^{>0J5&yhXON{HV-M19RKIj zO^dV)Gr?gUep%(ZtvNoV z8=X_+l;Furadx|@yVdUp8UdXc*Zghw$R8*!vC)EBp!ql=Vv^RK#|8Q4qqsnQrT*?i ZDj%dU_xv7i^r0b%O=q9fH2?qr005DfcH#g4 literal 0 HcmV?d00001 diff --git a/2024/06/10/startup-data-health-starts-with-healthy-event-tracking/kafka-inspired-startup-data-pipelines_hu153a73840b0658eeafb17cde694e5f88_121924_1080x0_resize_q75_h2_box_2.webp b/2024/06/10/startup-data-health-starts-with-healthy-event-tracking/kafka-inspired-startup-data-pipelines_hu153a73840b0658eeafb17cde694e5f88_121924_1080x0_resize_q75_h2_box_2.webp new file mode 100644 index 0000000000000000000000000000000000000000..054a9e51e1901a38fa0b15bf7dae59a52d05f13d GIT binary patch literal 94984 zcmV(tK)qh8|Hl5Kz71v zY#o2t><`V{qB)QA&-EYHd!~2;c+d5(_P(m0@3?|GANt>559pp#A8P-P=qK{8?mwwL zLOle3K>p3_0s6`5v-|HHU%3Cl^&5qaz_yz1FZ;X?upi(3->t?|Jx8FQzW3Ju?d)s) zZ=|wSei zy!$`?Kk~k>|Hc2)(r@Me;(Gyqmh;>Hf0IAmzw`d-d04%+^i4F@{NDh2`;5x0zr$m%}$Tc4lybK|=pz3968>;k&bbRVOPF^5Fy0V7Qsljdb5O+0z{AHatqzk3Vos>(P3^e)oyC3FHbss1an@gZ8@QBVY}lS#Yoz5(sQZP{72(TS=>q&V`T zuAhM$ybqtj@)zPq!t2#t<)&G7 zo+~qj5_T|moYbT>ly|Ztik0@`HrJL#gsLSdbV%;ndq|`-#`%Q(WU(+@R$Y_39rDRJ z{8Muw$2J#4WMx#sZNDabwY>*1T_4U}TPp2(-SAb#tpIFt-0jbj1t_i_P}*{#U;r_m z4SF-ym3N_U6`Wpk3>6&eaS(_FsePqgpaGY&{9fXml-Xi)nrwg3Trxg4rwf&*8VpE+6gG4k zfL4jT3v`V!u6_$?z7e9+P9ARx?pNMZmmjq`@Hm|wOy6*DApeO*+w>%?g!+w|)-SL` z6l?t=xYpRGx0O;aVTsE3_UrH?{OW0ggOm@2)pA_SbZ{pP%mX(TQ-j@{Xk-6>MmCqY&Vy!O$?K>e0Tk3VTWlbgE!L`@85= zdh!}2Z29y2CM>i&(K6buEOEeElUMZeJ^B}q1Kg?^-By!;KXy_!kkeOOSw^Ces4xn_Du6r$#Xk2uM$S6`Qs4B(PiXwgUf$8;A5 z+-hWly|?*d+vsq`lBeC3fBBw0u%97@|O%uks4?Hv}+Az~i)e zWy;rQ2CECHM0z}JYaqL{CC?NUtMmn>-AIU$FpAI?YoI4XC)%~x`esPOw)~sI#B3it zez&#N{!TJ#6#3tiluPqYj} zs_=Nve$w=X)T*Kg55iAwb&J73HkY6aZDLD2w}j&fMdY7~o-m>_>*??X=R|>g3}w8- z!joq<0TNJ9Th}!2A!VAE9Y6msDDw{Znw_xaU!Yd|?|O)L_;J!o;Q{LgAn7FMvJv(a zFZ$tFRuZF?2@(Js#cIdo!4@u?OxbidewT8NP?=neE?qq<4vgIFPgtYQ!n_@qZ!biI z+Uhpx0Z&`n=vE|i<`D`mg?6*OuRNm+!Aii0AVD>eML2?<=iToQ@3ZLCPA zCgeuff|Zo;kocuQ!ToEx={bHp2#?g&xR4^Jg~5~0T^TwgaK59u5VT4T)*KNP=x8+u zuF#I^JLi59Gdm-2bQFz12i$A!inA3@!z`pP&V;_D?4hC{G=FE+x##t8s+4=(Q$nC=#e7v#RYXYBM1`^aCUck% zcn9MfI!S3lDso}|lPq~JR5j3g@EV)C-Ts6isYqL?AT$E{dbb@~{GF!~)23|@A~@)u zgK3U0GTTOB5mhxT$v$eWOgr=I@E1PdTTH1O4P`7lLUQ2q4%(av zvRaY_841&87B(DeT6H=1RPo?f!% zC?3+2Z|#uLQQ2)Y2r=VvIog({#}k%j^EbvnLC;wefqoEP6u3Nt^}FSC12e8!J$~p2 zDTZsD`ws6oIH=avVEUPMl+^{rA>3ft{|Wj({-PQRGKl$=J!KXi(VovLr`Thwnr>?1 zI;6xq6>~$8FffPhv^NM8ebFB>&lpKgAD>8_`YDU!!RYI0g~Ld=Kbrf5dkgtdVsVlN zIHuKC+hBLG|C-U9>EVYy#5%Q|`ukgZ2!^K~qp3Cex$eHWNL=C?U7*Y?3YhYDZVCWF z9?I16%X8`?QWRx1sxqlH0}Vn0S!oYp`DgO^i^HFv+}h-BnHE*SP_tiS6D<%#Mn8j~tP4HV?SV z(j2P06lQQb>FZG=tEHyZ#c01M%mj=iuLZyw{&qVl{9a%Z4E~e$QhJEPI;=G@A^6?K z3E}M*^5-xvvOEXsM?g#Ro?U<(Ou?TEq$vF)w|ehIEnuNd#Vi=={iqVxh;UR9y)mq|LSIt&c-AtLOajk zylAi;%k|rcD0XbGlFZMhE2ZN$5oYP?g%W>Byy!rCn#=s1%Qw7vLpP0PE|~!Q!8b%O zgR)m2b2OYmAk*f z;Ge%DDnv(P3c^<8E-oX(@N)=$MpA=;Ws&%$MGXuv!4V;$q9_UUACKMNMi{D8BH#&X z!rE5) zM(!|$Mfo2f85WSF0Fgdnakvj~&AgFgkrc~0Ah&e*VYzVc2XKEcsCC6rmZI$nVV?@N zXlq}JAN9#Jc{OG^`OtdzkAc{S!i&2a&SWDsN#A5$)5B5<%#a0X=KtAj@^_YTg;*f_ z&?4hXfZm4n1Em1~be$Wr?tRcl3F;N;3OpGhc#T1sk;bXI8ktImOfV7`bZ zE0S$NHyhypM|s##8g)Rz%e&P9w~FZqp1-V^dQ@sSEW5oA@FXP`s2H7!;iR)q6{t?5 zqno;hX$08qA4`Y6ex1*^XNPl)119Gbs z6czd1_@_=S7IHCFxMPz**zoIBw9$B#pOPa4zYXC1O%b1`%Xz`KYgk1+79jt_i84q7 zvN6Y!&C&FSq~^gxl3osaP|(EgsAvmNwSy_1$?Eg0-$fBPCpxWK?+m%#e>PU z8(KSiz%!FHgCWQGsemeZhrcA_6K_tq#vjnoWa!f8Ts*)G34GR1F6=CcLshQ&I?M5w zBCqd+D9H?{`>OOb!)7R^h>U_MUQ*#9f(%^LQ~Xl7pUnr~9QmOsX4;q98)VF+Hvo=i z?TyBv+UAJ_uCoOH*A7GPClvAZWRi|m+NECP zNT>{uW|pRLQ$KA)v&d!BuT_pP zFeFr;UIrZhxQ?VRzjWJg(Y+XImMsC<21epYVVON1>+9zmd!9zud7zY zj4UnGCHS`wXzie`BzADqCI`S{;ZLe@%RGU6ja)MwR6Q0jvaS;2&g;7t9qu=!P7ZrU z;!x6_T)B`nB=@PmM3-q$dJ`sN?F@MNQGY z^O9PM$L%@Ngo=%vDv%va1@UoYUYaW#*{pxSuGJ!c9x}h z-VaxDN}feS*Mg@S>}4hp^PaNr%m4}QcvM^jQ%K)H7*a>jxPRinJ^s-lvU<1n!=;@AmbLcskSG1*L<2Ies}U$A&&b@|8mJNZoMV$914LiF1m`3p?FU&=sOHqr?D?8?iSLizanaTv zIn*GnfWIW-fN%~G+KtXI18RAh7`XeVwD&JOU{@wjn8I_B5%4_|JsvNLIH)wQp-Y;5 zDsL&K;X*!qV<+@Q&DW|6*J`)$;hKdT$YIHo4?)xWcEzkG&M~q5)_(l^`JK1qh<`@KLbYq)-2=|0^L&Kd(iqcayKSi_y=Ul}FJOS?EWxx4#w_Lpp zCWsL-92yWAeV!~YB|k+sq_Od+7%^vGB)5^JAO4ey&zO8<>Y$YqVyGRKx-}a;*Ss~G z&z=8)S|UKg#5Jnna1{;Z<599&U&1R0GM#n;WJrT6Qb@_ZhD_gk^^(98|K9+WSoKhF zU1)k0>SG)80fS=ymP~3F(<)XsymY=!}9o@o_ghlh$AU_G)}!Y}h>W z&w%G(&DbUVrlfbl9ty_eTgz^Z^_88*<;A6a9r=55Qns{>*8&^k z-KL6()dfVPbJ^&M8*?tZg`@ENl6^d>j!~*tNg3*$2KtNq?@UjGUlE4S@*Ofx$v(kM zV+%hl9V}?D%e#+FV`*+-nV2$si)@l+5W)(gvs8=>4>Yk6x2N|ARTKx>XF;317eMF| zdzsNYGCwD*dOoIF)N6*UC>VuMXQl4xLj(sO;ug6!Fj;!-LF(hO4gvCrb)P;`CQH;d zj`4LlB9oX~dF0_Oo{)voKbBTDc%S#1HAv|T;#>d{GZ8C*~s=w`*w*f=b?+ zQhq=^%hb$M&!l5G$)*At7w_x?RoD^maU>)c-k%M6N39>sX9f82ixOC2NK@?r4~x~ubGCDbf2LYw=!L)H9d4l7M&qbDwLZag%%FdkD)Z7q+BbQAsp zCRIg`Y`MS7+e+Y5rI}6G7xhPiwkc7GXKdk=_;@_6a#8RC$s4{GU4_1_88!zFoke{A zL+e#awnvCzw=M=$nNYaiZVsk=sq|=D`I-eB}OS(ZNEJ&a6hIB=zHHO!ugd2o%qiSbluB$M&NclinOdk z7uSCs^X;2_!hLUdwo%Q2P}%%*{{-`gal0WQT3nwr3&%F>X@kHsB$*h;am+xim*>@^_|$_{l6W6P9!_+02RDZQUJ1a*36B4 z9)gR&F*Gn{HH%;@8^(J;M?&urANlVz4W_o4%xlVbAxIc)*=}!Q-wg@dn`7H29BHJy zCVTPuK04%iPo1j#*W{JQ@?tvRM@r9S?|;pc`l1rdz;FoOPXqfRz`)?b|y~Qv#o^*Y}is2SWFj9-rlcz2%*h3J1Tx4+39sJ$Wtc`8a zJ&uuJx%1H`XCTdP@RpZrWr&d^IVor77WdVcRxV>SyAx^d%sOG>!(~+PrL%(g(1<`crM~KYr?5R}*iK zC^`fCqJBkSy`o6uJ-yTHX>=4-rWD|VO>`b^4wW0G!E1@4Zo4M+RF6O(xn=GA z5)yQox$)q`UpanxUBQcHo2R`zcO8PJaV(qBAAfdvW^=NV99*$D>)<)!kKnVmC(e z62HdsY5Svr1gGebwfHH?PPP>@()#-*F`P&mdC>+zngPoO)k+702dehL)4ojVFY;e zdx~#yEAA9PuK{}-PeB?lzz?cUQu#O)9AOcu4LBSGVLx4})F?aK$U;xYfU)h3Z8Cwi zs2#3t`Q2AS$(b0YUo}pDFaBAh+#6?|DOiOBkv1}T?vt{eTFjSRwuHjGO9Awi^i?qx z9zUu1lB^)l(R_}<1)jn8bv>au;-gI3m&X*&0s#TtuPuc`95Ft#*b_*fGEUf&|C{ZG zEX;1VV(JTnzdd^`#HMvleUC8#WcF{vqKwb9%nBuIZH~Gg^Zn++n-|6oUJH-#@Mf$+ zHRH09_Cw&{KQQULLw3R(g7vK+3o5Uh(b5-UC?}5ftdc*GG^^Y2HRWmTw92s5xsR-g zXRl1;R46x(I4m(j&%3@L|3Yq}y%9NyIpWR9&T84JXi^1~S7r>OE3EtXn!HA-!nOd9 z`3Fs4T7|EK3)xUWT)VHk^oa#v3&xVLnE>HsGo7LRxT13d{g0l;m_VNcTlL#{N2ZKONue1uC%=jY^2bIuJ<#4~?+{wOM za`6xCMkB+x-R~WW$Ec5|3{?97_K_}?ZJ~ulophy4V`5Bm03>-Z$?^o(M_Gz))g(QZX$yd)*AC;w&@$XE%U5)#hR;*r#iv>MrNuV|vOWe278KU-f_X1D z4uP*NtvZ0cb-Yy{5tqCZ+9(sfP*OgSx5+!)zT}xsQCc*!<$=|0rwiw-&U?OD#pai#UgWJvvo+(XDx8u4*RD#lSX zD=0uJv*Y=SW3yiyx(@Hkkp0WE;y|v|g*wt*;AOzREkhmGNk>|gl?!e5?dK{)IP{fY zJ53E|12{Or-VaxaD1>R}zvF$$<5<$2ikOqxFKyFc$n2cn!S_IKyPF}(?%o|hSL7u_s5=XI1_&~bAjyA$S&-~i0QFbBKbV(T(BZ&1^xomzdyE7Hx` z!t;}@3PPVJBbp6|rpVbz*Xsj>$Rg4uuuLscgubCzBOY1PBVI8vA|U3K47;&Cb_?SG z{fl{@cDD_x{Ym!J0)@!8NgyI^KpbW|-f{Ko~htL{vCU=Y@zx+k}LjiulXJdJaFv zfWEA~k?`*9ikxI>+SbIRV?H>l@Hs9_?4keCp}bH-(uu8hR$I^^tPJco?3*vTs=e&} z3@_GBA;z&K~z*KZ$4u-k+gJlft0y{;4>XVd^ouZyic=EfOX1?uIq;WT7~rd1?5g%v!#&b1@M7(Q?v8i=r;2iaSpuyY$u8$ zxVn`_XFcbpk#PUL`9TMzW?enOl9{476h`4{s#qxygl$Y2NgCnTqFYiqb9Pjb_!e1I zjNyO#p)*1bFY)XCg8VC}`5hN>k!A3PX&f5Ls64lGSJKCy#3cBM|0@Ax6@T6um}pm| zC=%H@LkD~G5_THga-EJPgd?7weFEsJwLPW_@ zN^LOET5!BN&sOp)&H#k`!3)~dLrT*zWNYY|2P~~f?s=Zv2!R>o{32XhD1SfT- z>KXw(&RHR2o|jN|;_;VQi0&#_q2TeVO5jnyf%`o*aAsTtC? zcApaTtv5D7a;I+BOdljzw!#0WM-PV)cO_cd<>Nf5h+!hnB6LpVN;R>r<_V$oJni;~ zh*Byz$X`NwGBMr*OYI|SR_NA^8|lFI)~){QS{{_CfE|f(LCq6a1q5sSBCr%#tkY#g z>~R=2qZ51^2DVJklRE_4{`Jdv0zY3oOof`3*RXju&1(sD{bEG>3t@yp6+Ya&3lv2>VcDm_!`VcvJ)qbzwed!mRyC7W4%XN5;B6yL8IxH zIj3AeKLoyb;g@nVpxf_6q?4y?$_@=D+@NJfjzc^8(wp0j26xYj`!Ek6a}xY97v8e| zRTs1tJJLS6CM#I+jLLScK=_{P{h2hf(w#m1X4mo&oz`fp{h!~vcVcRfDz?ijz+x+Y zrD|DoomhcfZ0o+Zy_X)58x%OBcc0`jwlV*s8g0m{8<)l(Ww(Lbw(4If*mF`jrniFU z&01j2wXL-XJ(k)$<+zzDo;5SxJiMD1f2`E&HD9Ev{Os7rHBVy;$y??_2{sb#Cr$)R z*FRb&#mfRG1%KO54*WxK)k|^L@x^ju5?9{-(Ib6Ng^p^24gX|h z1N)CTC2z%qT7kR8lGP~xXaUO|M0~RX&zCFikX}Wf#3r9PCDdC&8G3|cc1?E~>&Opd zs+cXd000n#O$^1!h!HVMI5DxeU9#z(7;P*5i~+{CqakN%G8sD?Z=y0LvXO1FZH2H$EE z32?}fdVs+_#H%|8ay4znuH#{2A5gzeO1N(Gt{~iGF+ufX&Jd;44*{F7L^dD((_8nL zJMmk0!rmS21;CqbBK{X~#$X}u=^a1odGKyymka1L_;C^!rk;NP{vA7v>;7d3^A)K} z&`y4tt2zc`zO8-> ziWG-_4K@X%*+=#>HHsnwR(8N*(|&G4IPNe15B3am*=pznDtg`Bc6vj%fy}1m)Sr5U^0)JH-;!OoSlM+I zEq8X(%h-+zKd23E>N>pr!0j1qzs`k0ij`Tt^Y^_#hieQg*J0+L0h79%|PLNtM=xT_DaUJmAt);+ZB zW4>>R4`;qotL<~^f3NcYYoyt~`3tjqbAO0FhXd*0`Hu4@%0ER;+$H~mEIRb2<*@Cm z25;{xYFK*AZybJ|c-qFi8PL?6rw2b~(kczs6HAf(arY{bgw*Uubakklp@?p}Ra*TS zsmr<5L!86c2w)}$!yCVlgU-&*DeRyFNR|8mwb*N10FRur6HLt zeU%ZX0If(`5`=-KA?Mb<8v*Qc6Q|H3(t>UkEuMniu~)%gYv}Pu_Fc^X=@*7wD?YZr zUav!Td)b5Emcxh8Cgdl(G5*mf8sF&C*F2^Cwef#NGjWb&XW#8~36UfmrQ%EB^_%~N zW#Ky0P2c0*o!c2!3AQRMc^1;*Na8=Gg(TA`a2%p}sDTuUIooonu z`lnZrrv&4rfD2V}8ZD`AFdk?8r(*7qZrm_h`9hQ%+!oXJP_Y~1a)>yl&z5AKY|Po$ z17@p1Oajv0{HQr-Vq2e$>nk26MK&}x#LNw4?A1!B26RdAqaC=Y=fwq_RN!FWZ@C*eU?}}X zm*^{mUo;l1ep1y;nSFUvaZ{0WJs%>c-R=XroRQZauy;R1X4f!qR5r!LE$vDL(;V1& zt<&}XluPM;@)53X+Ap+?0r!-6VSy?A0YSIsZ~R3Q!WW z+EQ}UOD7zyZd&jzt+k^?HCs+8>8yek!t%&eLEg`7WCHuk%&;B{(wP}8{PY_cch7gkh&6rD{H9H^|4 zvkB(JRTiYaNx7gc2V!33YJ;@vzm~24AQThVl_i{h;(?m0K}>yAyw6|@=g=7xtx4+N zU*^V5`Rq&#Zfy%qe)-IG39)ER%KSsR+f0|>R|g{IPP6gEVs`n3xhTiu!Ai}fOCS)~ z6IgB@8=OtM+mw@Vdd!d}v4_dZ*6+Nt8y#I{ymbZSb45J|=tJlFOZJscBFp09)3 z|E2Q(=j|}>`<<|>_vl#S0q5=P5SoD-YucyqeTh4t|L<|f$-m$G&Xf6n4Z%6Ma_0n2q7?_{x)*M@`bi{UHMKt9isDhJLUo^KGk@A2~+FV(eA}=XKSt+h+`_O_^W%{Dq(c&7=QoyRW)UH5f0=eIppZC!dmG+Bo`(3k{7<;)1Z{673~wQQaa#W3ZrRQX z`A_AbRHQt5God{j!p5|+XsZmvb2W>5GQs~aG( zb?P3x)AGi}Nb1Ua>!H+kAC8=?eU#K=rTRmU6#w>>4tX+uV+g8J8x3zm*lIC*I++_`|a)W0>-`{+L``^e^)hyj3JjzPTRr^lq@3^ zW&h=69U;2_YLF(1QaT;#?1ox^NSAzirRYe(C-!_0Ax539;f|we^5~khe5_^vt^r$~ zdjBNFK-P)_5}5Jc$yrk@8(mT%yS&cWp*2>N z7_c%Dg#IWJwwM<+j#VVNzfuJSzQ6mDw5K?!Z@w2MlLx-v=jCv8l?(M>?QLgjDPo@H zV+TCF>gC2ARO0`s@8VJ9p1&rSC%9V)xF-Gk!fox_{lm2!U^faSxK-SnlDv+1!IGV~ z(lJj7*!3Wdn2vlnx6x|sKBRLEf6KJ?0#x`#b91A6fQxeZDCztsO{K>N&R_u}N5d|x z%p7p(GlEfTufqKKnGH{mfVUn}ixf@bTNz#Q-y z89DiwyAWflwYF4HFXPeJpZ$T>CB;^)&8$GQkrLLhWo@eH=SxwRW}r9$hg|t;^zsX( zBPmY+RV!yNL^{;2MTG)nb`h4};^okp07gk&{ViZ45TS1lHa+NQb4KIyTHtyWUE~m1 zeZkvIkvpE`AJ-!XD>>#rT+uaSXDDYWTpbg$XD688?`;zc9GVFnmCC3#(HKOyw0(I} zc&pkOSlFoqHp;souz>qV)IP^aBGZfDhVY60Rxbey{WKDTBx1URLUS#-B(KswM5yjK zBK@XMi*zM|nx@SanRGGJDb{<=1W{>>`~M-q6Dk|XOGKiQKiKc;oId-q=_H}@`m#xH zSZ%yQm9bA&p{jAuw|JQ~^vJ`lYSA76G*?r~fMv-S8u% z#Gqxf$uJ8PMz}3Avp78Wv9Np%Z`_`}jOl7bAm&O3ORW&z|#Ew8CC>(VB-`{v^Kkuea{w5$$d3m%v%-|bFQPJkiO#M z7?!GQA%+fmn1C%{z?BG7SYVNrfg<28EVJf7CJGsra^S!_x)ex)(kRn~0Vnx5XY0MCHJ%*|y*41V;nk5iqy#Rg?1BnSH6a8Wh&3=)gT?>8)9=Dh7i_k1 z6z{M@Y$s?};rzg8fkxg5eN!4hKIguFw46Mu(83zp6@5RcRlB6GLs<%#?ZHLMw0RRN zyej@`UJDIUJK(kB+D5N)>AYhAQQZ?cQ2x`Eyag>OmbEiHr)tWdy#4v9#D0O(!DXCz$*Il~?l)r+qkcFj4sF8sgXKevhDW6@ zw}xO_6Zrt5lo!_T&?=d}Ixx-2c)p8ipY@WivAATpbxs~f2b(IJ$ZB=*)Cx{0(<5~B z?yWsSL`nYZG;xmaUsdreKML&dul}8Xk?~289~^@lyMHE<$U=Kq`=pqt*s~B+_+l3+ zcfm6qNnCPgvdwG)8jx67A-_xS2(J1Y!PbT9Kb6j4Yz!}Xg;t!3iC&nG-nC>uq}$EX zp$A77E=cZ&%I2};@WgK)KW5@ofCi2Hy)JurIr6gv?4q|sI9AorxOUoEg_{p1>q

    On?kx9b&GNqxN6j+u}) zhLJ;$##bo?yyii(VD=jgbzTUw#9tP+#y+HA4L6GeL_snhkd_=avDMhB#!XHcNqX722)BS zX1%$axXy-#1i3i^^92msbHQlDpZ}7KUxv+zow~=i@t*CU1~5h|ojl(#S>AszvWFbb zRP_kUIj1QK<#n8?>H+1OV~cX0PokQ&O&ftz>Yd~@1gb`8tCa6Dn!F29RQRkjeerF; zDYW~081`EXHKI!-DqrEk&(`W&(fV-pTFgyPYn&F^v(0+RfL=FIt;0q~H@zJ-*TJWP zg3PT1a9|27MhA!yFBg1?c3?YB_dD{TPgG{qe&u-_SefH%cFkA^!uNGSJ!V zqcQWLUH<@n>5(Cd4?}29bHdah94<|vyQ?yG3yBzMd}k)8nE^xI_>|J+rKV@Nznh#4 zlVQc7XheTy%)e+44BDf-Y*T=+x~aRT@upWE(~h&Bsy_ZtWf` z7&#nP7VNTI@v&EO>pyV8+LzhRGxcg@O`@L%$Npa=jmDRRNVzc1edZyC>)_m-YNGephqdkXBtODOWSKZpt>#gkZz`f2 z<$}fFDI_HneK;vUV$Br-u_VLr`A_B0O?Iz7zl};&B8+U+_ar|5fSwyX*s+tMCv5le zBz%bb?jYuR@pc}yKKATU?}Wr=foewQ-`+nT@8c|Ofhab$Lc1&zQl3schL-8G{MbW){)Z(xZ{bCzjAsb@xlOBC z7yu}6rbTK0VY8*4gVktct(BE5AKqS&>MQlopvjHx*4$@nn=giIOR0MpisG7oYWX|y zw1rt6>ZuhitZcG}7>z+Bq><1sA5rXonEXL2uU**^Uk_q5$-$#fil+l|R-X{^qj7t7 zwv7Xq-z@SG@pKYaSq<7U?uLg`=nW04rHdzB?Q5M=toTbR0^sXjAbQ`&$FN>{sEMQrppKMDu(mVQ!miXCr$_S2SC00O3RVFG6tFvX$qza5UBh|+s1GOYAj7}FTr?* zn+iT2&x2wMJ`=a0;}eEqii^rHtRGV8sHd-aqMBd$hB-re8&>)VvDwOv1We?VkA5dP z`bo*lyV36$VH*pVVhU58p)t`5WurUn?zbNC`NqSt7khv^Pep;HP3UT9^=|vdLokyA zVMxo7h9EuXIyO@6A~T8~6h{{>xsv-gNnnJ*U(MH2Ca)0KNK>dOu?75=#)6HVso-H- z*VL{)Oee`i5Jwn}zVE5^{dssAc6OI<@gHi1#$_07sA@jaNf0t9 zCW*Sz=P!cA)8)Qk!vGa7D~F$zjSV?~Lh-1 z3Dr8rWNMG51@Ktg2PL-~2bfDw7|IcZ8@iskY%w^0Ao!c$CI~olOu>o-AGDbsl{eZd zL!V*1`jVTYDqvw7n~qSXzpHQSQo3>mh^bjI7#r{{%WYy5Q)zZVsG<(BZbfj}eP826 zhHll5bOG?o-=>(p7&dY=k(kQ^of4JHDZS;sVm9E2tK$>^22%XR0Ro1NOPEWvA%W~x zjxAv3wZ}JSnmy~ySBy8pKSED22-m{L2*T#HQrjDCFRD;K?bT51oh%cl0IL(n;FV-W zoBiA~)KHAahnW!@JtmLQ;}}dp_}R`mia8l}gXzVZ&O%s$_h8ITsBW0d`(T9lxx`(6 z&ms4O^s5kDUhNMziD_{U0sqF(TWqt>=m10;;sog|J4-mUgzOhvRQycAAH5mjL zrm8S}o80ya>JE!B29q!a2=*7)u;M_Y(*GX3GpMm#co>i}Hl5 zREW1yarzFhU09Y5fj|g&N`Z@@^1LgI)h3g5>o&t7@Fu7-Ve`XR*a^+Vyl+T`&d2L* zqkSUA%O3+P5WHf1h_6raU(OjzX(vv$aTffB+|`X-k}}e_DJ3Xkww}Va*#q)j9SIPq zJvl501BPa1{x}qI@5+ zyh?{;4OZg9*`QABfcU(Fmsl5^|MMCK&_zYx@d}a5aeU&>bLzQ;p#rH9thmSJ*UUM4U5ZJ4Z)Zo^3O>P7|Vi$D&)2SjmX}Z#BU^vNZh$=-T~86#^9M(rrTjheC`L&Y+KOC zc+@W~dAaUJ5OIX}&a4Nu=(?}ymiY0i8n0p+nC4Cr5wfum>x+cx;)5^wKI)Eo2r_p~ z-9oow=;){kW7DtMLM52w+TzN9niO!c6M<-Z&+5=_v4Ak%v0Jt_#mj!(CSDaqLU_ev z!Oj~|=@xDFt+YtnF3O=XWXJfqE}tKQ;jI?uVKi%28_Qdq$|Pipt0lNq>A%BG%i%8& zhUx*fBpWF~>C{xG6f-cwzYPIO4mQc|6R6U+NYv=+SZiIfai8G}wXS&i0iT0FP z=qkdOb_v3xm=Z$!67qR&AV*MPoCC%-SiTO%o| zK@D+OQ%+}oXS+hBGFmXObZK(|{O0IlXA3IgevtGa>7ym>aL$Yv%ONM8-xkpkPD~(! zARkq>ie(afL7tTR=?0vQUn(1_>TG4E7Yka#k3sR#mwtJ;rO${C&5N6nCx`-k zrRxrd%tom~48`GN4{(1=K0Otc#Sl$*}eOkfVUX{-Y8a?5}r+9Z;di{7BMq7ncpqb zH~f*GC(q+wE}svbV`%-%;@0#ad8z1`0qI;!t;h`tgfNV&ekg3AI~4 z^L#dzMT$I|M7M>g3JuJ>Uw)|63;4M?b=)7cxz!|~8EFYQNihTt|yDOX0J75?6*N zwul|b7g5#6;+T7UV$rEeQW8ua`0|0gyk)ew5JcId<I$N+bt zDUW5xNV9j}dZ45xSiAJGq7alsAGh`bo$o;Nsv~I&$+HNJvM%6raf%m6HJ#RS%f{i%fR@f6UYro>^BR-Elj?Hzjl(?P8v zPApv)!ylIuR8cug_!2zAqG@N_1){2|-Npf*;s0{1pbvfU=3(%JeyVEjqTrR6A^Xsw>Ox3;u7>Yw=0Zapm2A7;rW#`nt@dcq37bBZ`w1V4wOP<%h-20SG<_Vz zZA3^kS(Mn3AI~m|SAVVrIhiLIZ z-L*2x3B~47mI60c8uaF-1sc)?tXgimh%}T04vpq$PfE>p>4a#F85QwuD!`Rh zBuT>Hw>K-hgWer+aqtb(-}wrZ@1^V~De&suGtkvRGy_@irwzf*8RI%&`8cd^`aR;) zf|N6%)zlg*uBqLA^qGxe^u(>c+(TR%<)n6r7*~EzS^yo{{4nU zeLDa9y~w)%^Gn!jierjw%`5Exq>yhn0KPT*OTVQ_gc7e(RGn7FW*OYlLe9ckRjpKx zh~Uthvh-~F&M!PuWg!0L@`~mOj5bFm2Vl)rXC`F$(yW5Zm^ut>MzdH_3h4E$(+IWf zX{AGDMCOHYQQ_XsAZ)6NTrw3yPRwzA69!^xg9B6ZG!MwyRMGeCWN&G>Cx=pYir2NAl)L8~7KmBCUK$H3oc;pTVhzpeI0wcM)I+(`8Ub_u*0?bh&joj1$Zv~BGyxb|yPOC8w{kCaI0$=(jn zl+H$9y_m!Fyi}3al_8I4tNb^w8rrPV1YSWCeb1K{o;jQrqWHc0xCApKsALGJ*N;jY z#OXx)2eJFWW7;`JKQg>`6|3C&@3mW$j-43bcpEln;mL!g=gq9Yk}Vyz`N+3A>y=oB)vSkmy`TpIfwGLyD?J4oS~mzyAeK9hY1S` zmjO3-t=Q;fuR}v45m~0gQAVHx??;1K_r;IecYhk&cQGx|NkBk6E4wBrb8mZpl_qlCHpqnBwue6u48b z#T7#oWf(@Iv!rOV*DBbY=I>>KYIDmR!09qeSwVcm!HIUn{Qfk~fngny<>D~4HZeXw z#w~T+GcVdYpL>^U_%kT64nj7d9X%x(OLe9{HCL>OS1N~Qsb{`gHsP0e2*ja~J$6mz z4*Z$iwu~z3e2B6GdD47l7>U;i+z3B;mr{Y5t04ZtzuT5{Ob z9S@lSl#9L89|-L$1RzbXemaLI$!gaAG3E~2U1@;U!=QXwt&-e(b7iaf^fcNJq^kIP zNEhF4iWCeI95-KiPDYx#L@5sguO9S7!OL?wQ>PsfNC88rBoj?N6YNWu_NM6(#VX^$ zTe41a>~g8^9+DHT&kcmsF_W2*&Nw6m&}f7cSpCvy;C=2pBurGtSg| z2?ZDH|Flts7%PN9RBSK=4;wFQF}1jqyf0)*?YCrykd|gHJ_|dsS~x6QtEp$^g82Ae zg4jJQ7{8HVxR~+Zq~aEGln-faQ-43@6Eb`PrukZ3z9hobJxXZ~;({W<*|K+GOe&K& z!k_e27`2gBz7Uh0;##O=#?qQo2P*@=2CID*7BNeLVRp~NTKB!FDR+X1-2+y(IQ;4laHhNIpn4G~qjS6ta}n??0TvfprcrVtL_A$gO%@XzGe1kc>W zK^0z9q-27X_TepbtGy(is$*EYqRs;Gu<|`j%yWkmfkF-F@;kqyZSTa7vUe(kN~xF# z?Di)q|`n(r87xl`&II4ALgJ#juEuA_O+24S&f3e_C4&d7X!#FOs^G# z80^_vO_LvKIF+|X3-=&~k)l8}m&?oDeP&gxtt0}akZu&LKFGfaq{A!+N6wWj%~MFw zyYHj-?vb+3dNjM`<775Wvatn%Dc7K~4BhuWTrrTb$Sc4$&Ki}~<_aI3gA@HO)z(Z* z_XAgETjZV6k=X)@JKQCr& z6gus_mqxqqQ~EG+IQZUQk}jH>Ry}HsF`l5{!#cVDXAFMifWt&8=QNb@=L*fHEE}iz zJ9(Zl;Z<8K@+79rtOaiR@32*_IW`dKGaV-tyg|Z?#vny$GkNW3RV>F{`7ID!^RVSj z^z}01K_qE97b*Y*pvFY4Jc$Njx4E;m@7ro@&B3@!ES9fnz*J7wt+idls9$k}-_}kK zkol&Lb^C_2U9yUC!4)Sy}vO7`oo&wu&I#JT}4}WU$>pyShX^ooQTkKUPxY+`TATpHMIG z{F{?m1W^fzqZ@I)3Sey|mJuP%;*fZ@NXMAE46R+9vTLK_PA7tZ>RH+SyDwS-Vb~e| zG+B$|5gSrcjA;f&(;&}J9dt!jJkU5daH8yhRAB+{N!$AXmVnW20o!nF0~wfEoAy%O zQZ<9~&g4_{6R3R9FaC|y^;-w7r(I>S{-o;Iz&w5#7ppwmcR-|doF4Vb0lQUcH?l$R z?nSaCWr5fPnD40lo-;2(dO!SB_s>CRLqRSMfy$ZXhfaPxIdsZf$Iz7z0N7Ait5bUJ z=j|$+aE&;7W3QbF<^U{{2fkuGH zsKYvr@l*gbJ2?EnL?+s4dI--3jI=4Y5Dp4bd}(s4uuK^Gc<(84;sF84~jauBPU> zFG!*r^aFfQ;uRDCTy9Q;ByitAHR{Th3lA>N)QEYbX4_0jd1h)8m3Q515#a%(XGPX~~O=3@Htk^qFlo5;0@glsZHw%X+JtqRReGCOFh# z0L!#xFCa;pyI|tK%XBaFC)XyJ^t(?uYrA!?aK44Pm>>tFjBhZ#!aewDZQ=vAN511& zb>jyvYM+-bhXNbcZ6lb!+UHWf^%rRNYd1nY`TgnmGTsR&yzue2&P&qVD#%ep=r5d| zIBIt`OkNcCp_Ek+U<`Cil&&(5Z_wA{mM2ajuCCy;5;I~Nl6DcDu#L{@Ny`A?X{Q9s zNbOtZX@v+t5!6H+B6W+XO|AG`Nq@X|lj6%JVt_|XeR`boL?ClUU6G>)jwSTp_}1e@ z_75WySVRYfu#`&TD3JSl!%xjN2ONHjj+Tt04)p1#|>WLFiialT?3Tje6}uKi@C4mH7i~S zmGv&cQo)&OG2J|RJPrX0mx6h#t-8IPy)CVclXWinXMsh|k<i>e15Hz;RWa}rZ2_}B1HwEZ81 z#KZFd9gLG+whQCaa20Jja8khCqq`Ncd0w+bS0)E&yx^p^+RxS0`1rWTS*RarkmaBH z@ewMKZ|oX#rVZ~-*Ds?F(sP?%3v}=h4IRs+4^u88pOMG=;(&e`7z^mSI(a}lK; zvM=qajpMXlk~bPI!jiPt81Iix%`#&1@$AF4HWQqe242^>`>vmtI;`^4$yM12cO?&> zJBZuzu_Dl)e}0psR_QLzplfDm4@u<>$-rAlO$qt6EqEQ~`JcT--PGg7d42_hsQO#q z($_b2&K_U=zDCfV6v+dVhARcp;x1*fT=%MYmjMdK`J%S|8 zn+~-9))21nS|VR@zEcN`cNkx2ZD#N`U*_A?hvC@`E-bEkLtPNP45lX1UpNmINA2y`;m?Uf;n zJ1twfmy;Ok_%`+pPBHrG7CH||Es^%Z5Adk~5?0Z)u?0bq>If6_BKm&CI3S-|(_wm_ z<<$2&k|sj!n)`}Pg?!sB2qo#D6$XKUd06XAbBStW_p3CsDo|AlX?RrotB)K&aE6yn zOmY;qyA4uAG8ab(iu2)7I~<(jnK>&zGo7h$=WrBO*~T7a2t51{JPs*X%1naTD?Ob` z!dN_$sTmJD+WNx{1R6W4OV z@&oH!h;~bI1X!K-4M#DAPr?B;UP|rD&h@hU`T`&5GUi8jfCXcl0Daht4L)>_Y*)Ie zsOcxN$bk}d?z4|gf?x-Voh|_Q%0@h#XiSiGH8Rac55foUjDMmcKNePunC@Hj`9H= z%=U-lB20-}*<*x!1#N8!YQ*5@Tzcf+P5ZY>7qNL6t$#Uvsa39_Bfc4?S9Wb3XT~MH z`^&>?dfqerRSgbY3q~PFo|DkW5%~9n$4$xxPAp4Jo8Xo^PAOci41`>F$-@7}5Y;C$ zcY@#s*b-_s(KT2d%iW282z7J2!2%oR@QQ6^o&GI4BUYr77D3Fi-hhOkXyA8SaD#-R zD5Xi6La3U#Gr zmB-v)QjV8Lnb)ZTL$sCUuM9>DB>`$YmZrrL%CJR!(ml3rJDlR7LoYIZiOBE*{9Q){ z*=)s!ZQ|y8QHX*tJD_kYQtTPbGfsB|)4rDW-Y>%P5PS?lkx{}q_zM{IFf%%7ZlRkm zc17l=5{TDB!twOGe}(%NpJH5$G5Kt(jyvZV;Vi$6O?zRFM8B*X$62=gdJkpg( z5UC9ewdg075u$w_0ng=O6u*V6H-y6n1qY}*m)?{;umSkO96VJx^YxA7k2NSiUuo)q zTR`5TauU;&+&|O!;c0yXl5Zhc?MWD=b3NB-v~5L>Erf{|V?JvW2|CpK8C6ARK`|j# zF|$ibw!B!HCQ=BP6PMm$ZkHO=`;N$DXM;aw>) z(*%VMLm3kO#a(;rg&=*S=a6}d70ie^6AAA2pE%A(rKm6Um#n@A(t0RV)vuKJz{%xS z@IVHHSIcP*!o)E~g@QjCoH27whPxDuI3h>Y$?~(qX}i=;k*RH-mqKFi1Pyv-unK^b z37o76v}Qd>^MYp%^%wI(eoB6w@pFFNQroOa>#=jb(!Qu2MJlP%7b{5JMTv{K-TTnp zDb&zUc19$A2&zu-z70C&bhTG7Va7gguIQKI!W#|DhB2zBxN8hg9?)`<#ucN9uRf-L zPTUbGB`aF_t~bU3Re!nUx;q2g$#u>|yk!qzOWP9@HgFKDveF)Zh2&dEEZjBG2iq>F zHbE-Pzgo=C%H7Pci(j6IXAPX$4?83Wq*y$Gbrz7DU)9vp<@@@{D{g3b@SABTqLVwg*0JSxeaD-<%>6 zipCp7tH&0CBq4M@j0CF+1~qg&NX-w}-I#fa+Y-0j*$~Em#buZFju)j=c2)&kWr$S<>BNZ%VqhlN?mwp~hRd z3@`QFxJw#aWozCmkT)1b(^(;JX#KCz}RD9Ktr?g7_{(&dk>vSe~^ zgpEhYcB;Yv^B@Y6!I}sO4NLFQc93Ss%0VT{SRMsL?dUlTm|aI4izKx;60OFf*y$HAhkl8o z$?nO8y5{teNk~_q_Q=lcpb$so5&ggqGD?_4~;u7EMW76`+c?#`%V&U$-2pN5mxj z=6cRoH77q~Q?9G>wygNDPFa7{d z>oXrY7A?hUp4MOxBX4>RkZ;nZJL#u;RiefV_Z~oyQ0i=RPb)OI6?!Y`8Wpwb(hsmy^&m9c)%Bj z@;nM%!z;l*r+`|BKdCX3scD&quU2qMboI+_SQXZxVV4}zXiFVz^(PSrfqT2^=MW7| z*RVQtZHuhUOATEjj+vAeD%Y;wif@Rm+?3Hc@H};F-)>|6RK~o%qtA(53KZs$&lb%; zf0O$qDIj2RsLxNuweT!!6xQ%@Hyq2g)%vjDD}awT=F?>yXt+fSf-dp|g3}h^1qakK zI=#6Qi-%u62dpMaOX*B`eFH~dB}m@1;3a&TcUjln8{3y6gLZS>sft{cExg9ik+m>r z-~~Aw887pn(R0GTv%lIX`|FZKT#_SKZJ=j<9F3!{qd|2k|S~cSkY~-q3o` ziRn&s3qYgu)q8WFn{UdWkW1MP=j)BRnQUt+AUB%~{c;G<9~^4~moYugMjnS(NvT}% z#TaMG#(2s7b!22fB##-Esu4;LwG<~dUyP=`LnKY3iswr>=YcqR5=vwjd4_c(nPAMN z*+U2bpBLD=_qCp4xI{Du7l}=&_3I4*Dcc@>aqwk*{_9SzGx_(KdnonfLs+s*=HoI= z=nF_z1qB%e;+rRNZ8b}>v7i}A7q~tK`eE`$k4&KQ(mTrXfO7*|BXije(8V}SL+YSK zWE`&|@?7hP;Rvp~h9gRRCJG|MHL#-dUs_t-8ZsrmU-P84q^(ix<*2_c<@iD-x$cRb zlcWi#Y7I$h;NF{$$~tS~7^sdO@P;ZLA=DnIn}QLSz=h#oG&QbHmrzCI3zU>9BJAYT z2I^L}g~oqI*tbj81{f`!si9LZd9uoNY08qxHi@N@+niilB^c9yOC+pc z+GUve2Afl^p|l-hW4cOsxLf=@rIS4Uom(ghElH$vlc}6W)|ip3-mxNo);D$BttYn+ z4#=gj6QjA1dp9dXoe_ElHa9e^`)0*!XvrrzvHAIy2x2)LHE-)bbPzeHv^A#eC1z7+ zUT5r{w0Q1{t%eMP5E&j2^Tet%5*4;zu+vqR6-W%k(X$`C_6t#QZ((KCDSb{K*+g-x z{@ro81L7~yah-EGP$G2ilrWp)UThPVhd_53jfSWL&{_LrfIh+6J{^T=c*9dFMH>@C z^$p}#tH>^&6vSLP>z8b@i_t{$pRnhIrS_sMXyN-B(Zx!&*WTV(dZH)NjJa}Bqom2by{p$848Kw7Uc{9V>(RTqq@}$N*3cJ z{>r#^nmOtY+G|UP4FONNE`E2CzYDwNeX-o){fH(3C)$yE_1+2pi)UafE+|sF2R{P_ zqdYiM1gDRGu00*j{*m_=Ah}1wigj3EjE2DP<6ajE7dWW^F(O!1zBJ9Me-uMC-!uOb zwr6WftUx{haOab$(}xMUor3PtP`(Au0Gw$tH0r`DOB!npFM53miBfiSU@RK(=lI`et*vx#>9 zq`^eIrg))-=r{YPiTSL-!w%l~*)>MW0&)~qSPa~%lp#EWFsWc`V%;b)BH>kvMvIr^ zRDHwJRUto!q-}wRg`~a$c4N^QN-_27j%V8P{4n~QY=jN+PN1YfOx4S`=&;>zA8*ny zjcDx~I&g?pG3i{hZAyt_R_z%kQ&d&3>YbjX&~g8TbM7WVzvhf4?eUWDR`YQ9QLd!R zE(sRHB)GP8X&I7@uGSmJ*FrRe@n`Tw9P>2xY10*qmv_vkne@t0)5XM zN@)OKnj5HDc|)WyjK!KNXyMl-U^M0a!;>3pg8U0jN;2v$FCfLb3a1a(&D9%prcCve zvzVTE-7(R*DD*e1o>E0YZs${q4&A&-Qc0yv%{5KD5Vj3VQC^d=L&WlL9-B{nfigWb zln-T9@wKPici8SShp&6oiCGbz|oqIvfn> z_x|LJj?ZS@(a+bp!fC`C$O0i`{_1vbK}cBYTb;w~=8#hJb6&p6)CqM}-#D@^FY6}F z^Wh>-IHg!cJgU+S?@XR)HmyQz@5QY(0tNdpcUwb=J{8nA(hu19F`Y4B2N?Snv%_=xw8EKFnq?^OMUuMROOXa zA2QY;gok2Ew{O7RtE^OD<|19RBdC1)59JDH=v_fu%!hG^`gL@g!H2u>LDDH`H2BJz zgj@LRp&OHo1M3KIVG455gakc5HsM5CubQJdf+Vw5+tsykqBG(Nu19HuD$bf$Klgy5 zHZa;cP$5rIiQKM;2E6Nxvl`WVAl%2>vQht%hgd8KM8A^XSOW=<*WvbJ!sEhW>qSd; zC&bm$Uk095lM^aqV~!s;HPVs?k4hQ^2Q5j$&f5~cv>9KQJe2^dPpotz>4R>s?Q_%C zhEeC2Sun6~_DO9A>Jwqh=G;Au0DN@GV4Ct>yWj=#?$W_rr2XiTK*>A7v>e&ZXmF=( zvUgr1Gjzz`KO~uj7Jow?en6D4@+;V`9#1j}L#ItEyT&^v*TlcCh1IVOO0#tFPhN|r z7x6;q7*DaQWNr_SAb+i*Wig43x!BSL8Rhmu=Ac7+&pExfJh%t}6lgWQ44uFVqS0Y; zxHtY?Ty6bh$qvQ-3N*k+tBjLHPmXEul9U1D&-EF{Ld6mV53zy}`!;* z%bb?M2HO0Oy+3ud3Rj`YLG3^|EwiZo#%Wdu7sA2XDw@`wbW<_lQ68K-D$1y2n9?J; z3-yj3E#{k&XYn1Cn?cgQZFd7F^ne_6h$!cmR`#OCU_j`^k8t*e zX>s*q>H$&8|JqxFz$`845WK=s&ZB`5XG?KA*hdhu{fHh9BAd(^CGbW_&s9@YrwMGC zvOAn|fa>On-Aw(&EnFUDPd`fb{kR)r9d}ZcTx=w3*fe4x!yr1!Dpsr)J>!$­Ch z<%tF{B&tA$Uq7IOaK%OohwWVz6}p7}0&l%tNaDco0q<3F-fYgh+vM0yLBpaZ|R4c~3a; z!3MrxAlK%>v{y_aSOIL27s^!NJs<5TOW@}k-(dA*U&^i;8)(uf_%p3XbOMuJb!WT- zdJ-e%`3+C>2{6<9t!1}QM)qE0Jsz0!V$S>M+!k9(4oRU&0%mL`KFZnR>AS3S+Wh4> zKRU6=AU(tgJ8GJ8Y@XiIs?@57UPk8m2Af+ba95C%C(wdY8Hho_c^328gW$tx$3Sa*YaF6C_L-ovkef1f`hYz>G<-5viqIQ!Mlb!8Wl3Z#1 z1z~0>Rn7(pju!F%{Bn=*ld4d{Q+RX_AN@%&^hyP`BW(4Nq=a}e;H$ty!m%QC=qzR+ z&O;)W#{cLVb*oy{y;epg7VyumIDDC>?31Se6 zRG_((&h~h69#N61)?G1mBe8jwdE~W6JPOpx^cK*peL%K`NX-XhV7&C*$I?B-Z@GOh zX?Nvm4}3!ij;3MO{9h3}vAfn@{IYQTXQ*=nlMW_+4K3;qx)cGl*}3L($&8e0yLUdx z6B2s@x@TDAJ|k{Q)xp|p)PLo-AI*I?7<0>cC25|kl0=SqH=qrAo|A_Bmn2v60(mSo zSa@)z2Hp^_$2}=wEGDPHdTzp{<1m3iJE&87kiRV-wE#JvY=)Xm`Iy4X8V`ZUi6-l_ z*!U&0y85s$SytYNM0rGEetSmTzm?9`B*;5Iq}f|L>Rfe;AP;{eg&c$PJ|U7d_bHLm zJ&8E@T;`3W$+$O0hi2bvNm9f;u{?=qi$?3hwkjzH@k7*DRWv-OvbJR>d^|*g?Q-m4 zxqumXZmaBRXu8}JSB~jy;mBLGSxb1V7Aq0Q+f^RZk-l3RnhYA9rNId=vbJdZl3R6r zGm@z?xNZC}nTHAg55O~s|UuQ8@(^|ks$aWaPL5~1OZ-PNU6-q#85PKc9}fKQ>p!G?@|QU zugqWomj;04opn~hNT13ajt*})l198(wQQws?u}(cg~5uVMJ}r;fb|_~#^v|f9X}3e zveKrdVax){XUN*YD5w!Ol7RWJb2t!-(c8O#Sx%C&r|KtSu|gBx#BbLLSGS9oN`{Di z#IuW*$)p*f%qe}zCo&}*J=^1w_LgD)zO>%2P-rdp$Dq|K@8M;$Q=sG#q}KT+Ex7S-KUmZKISIg$uCilPKBmp!bA*iNMle46{ehtoT_4U*)8O}{!hlpuZl{E+!A=5E_^hMr%Ik1niyU9 zx$AN_nz^XlzO+l$10uc?Ds(piBP=L@bt3XLPDpTLdNvQ~L1`X@74qp8jzeI-YDEjU z>Pi~k6;HnzG2I9RoY_EUg#V59CRTOy@C@hB+(*~AWl%iVl-#0SeAEX%Ldte~!PE;@ zwgl++>3%F)Vl;YNrgU>5_pH-aQjOBMz+Hde9Q95UOxR0$@HV_z>{%m^p?o0KPBg>v z0X6ue(gz%4;3vwB|5xeZACKfz``Zld!hS5HZKY!>3SLW5wm=Mll9Abxr|bLw#R8)4=F7Mq)%5ALjcddjcH0tYD|3{$Q}c&H}FdQoGJi;oGBgwL%}_{1M;Ac z_f7KMHZz^X4_c7yj=pb!n0U zTtZ@B{PH4V8f7BLzaY9aMdYO$0ZcqxrBC_e1ljIam*nPX+SmL5X_jN--QHEDs-C&6 z!J$4Zhy~!OB)orACcph15viZr0j{0htFW4w?ntq0S@(kzv3T#6x{eP{?gI9BnH5O@ z_sv6p9)` znJ!`JgjpwbN*L}DG1>x@7(^lY)aT1h^qB#Ey4Cqwg`?W;?onz;CR>&LK`2qCf7UYAawDSrqe+YR6{XIv3@?xKD9=G3@^lE@nXlSI z85s%(Oc(2$On!Ibh6i^E`hwX2x1qK(cgJzWt}utK2EEO`agJBwO(CKM?cuOq7pGiM?lw^84uwjLK?*hcsFDO8= zilkjPEsyh)n>_vNXt4N`!!P(VLNpOvtJNy$O)4;puQt})d#YpEzJQP4fi)ePw8WN> zxtzTKboOk><^r-GK7z6vzFM(Ps)Ls#&>o4ijtNTBlyo+>=8@9B(oun%91?*Ek|Mx3 zMoUvh!dv`%4*uVe1<$|FUu~Nz)cQBNpk!4jv5Ve!GXxf9Ql{?a$h#*Cd!;;iFnACG zV0n?_Q}?{Wo6aTMPL^$}5sAiz%lZt0L~3Ai*zh3@d~~ZxGa3Gv0qR~2N{d(^#S;*` zi8JR%Mymx{;vlLyBTal>U$rOXCkOjp)=aL*ft-(C{hUS?GGy*-JRl zk(aR^4S?VQ>E^A&Un5D*GCn_t;~Z-Xw!@f4pi)HopKA z6-lIQTWxd=+>mn`7uc5-M0B*jW6TS-#I6{P=XIV>t{|oCDd?<<TbwFnyo4R= z_gj!s4Ezx@iv-Nil9Jl@@_$0i zy2q>Rq}IiQ+nlz&$VRVPoWO+{>sC}vcLOBx`cwESkq!D0%=Md+$$~Xg_%rS8>{xy9=(x5q3pVBgW z|I#0TGcj>j&vOeMH)$B8N}&n!AeFMDXv9$nB+<5Kx`s;EWu~`xLW{L6vm$SBMZOSe zh@h_A88Fey)yPn#h7?Z^Y0YkR@-x}sRD9xunH`pBghI8pM5~5E`<}9Rd$lp_Kpcts z0M9f6yf~oysjR!PW|lQj${W!3Yiw>_e4Xuza|`L5OWvZFfW?cun=$I=5`mN>;{^O~ zFCuDWwv9j@HS_sb4R_1p&zCzSs)D)d5KzaJ1MkpTxVS$@L3G0=>AuX9a-Y|=(G-Or zjIIwk<1jHQq9r6ZPihTABn_zXMScLdIoJ8?w9Hf7^Nj`BwY=8%#G`~rLJ8#r8!wcy zx?mXDRIwX&WqYlQTC7nhiC=#O=~+wtn6_@R#S|~t@!T}L+g3mmnV8U7nu%)9uJE-F zmZdpI9l^c$dv9BbSnOaQxA9_p&VK&`LynTI350X@ft1S-e^VNtVbp{F<_V*rxY9dpY61y5OIOQ%IDubhakXMn zsq^rcEfi2L@R8Z`e*VlH|9W!rD8U(a4Nnd+f$V5)$(BjMP;ygCZ<@(+s3|||&eox1 z$9G4&MkKhGP;-w33D129_^za;b}L6zfG$1SzX5vcjp=8WcM-&J*qLBgLmwyVtzf+< zN8!-Qa%%*gJc6ib(C%9=s!y|Zb!WvV1 zc;&;Qs77Nx7zf_F#wB(^y?mPKgJbmrDjYgatnM!)_Ze0p4mlLC;XN;{rP&vFO6bMI z@3v6W?u8{6idG)AJ+iD&0lS6#!ZVVUF?3Hcp@h<>E%$o}CHGDKX=6i|i4TSHsZ`eX zx(1N|CEd_fLmB%ZHuqwiRd$>fcL%n-ZFuGtpui(Jq>=FZ=W&qSIsCyDV!Ld(D`$3e z@c+3uxAncjv@GduYLka4r>qIJIh-uGP4O}*tcNM=Y9bUNoSB|G2mt7)@e~Mu&b|;^ z=kNjq=TfW@rG)@TE!nJeeTDa&9?u$tB5{U7G5ml9k~*3z>yRq~LR*yNo8BZ1a8PrA zNuc5kCLe7$R2WYJUi|LY{9C6P7PuHzrd0I?=(dWOz(tDjlXC^h8)Ty-316f8+zEUO z0^gi;WoxCm!eV+3`S9v3CBm3Uy&+5vtc9$slI|j{{3q8@AzcYFGqrJ34C2(VNP?#} zN_bRA!fa+&7=Ef1$t8klf=6|{HE&H6k{RSCtYSZ}8`xq*#wy&zgz2|Doxc>|_ZDcn zvxr7*BbJa?n;zuShL04RLQzyOf=Qb#!UHk3hc<)vFD~x*$8J=C#elkv?t^fCChr(7 z5u6mEAT<`2b<9Mg*m%sD5(Tj-Y)ru`f6-fg=$55B&}_WbMd3-u%V! z_n>Gz2dp(Y8a&E5J@S0b3lix`ppFnqf4yUsF%f5Qo7VQSGrz!~)-zVKX`#EuW*r2h z>{>etx;UJP=WW*$|4DFNMkC&I?fw)ezF4V*$$OPBIDQjcOYf;?kr%qGu%%^iqd=@1(aP=kyb8 z+-Fc7wMc(VYJA8b6k8p{&;zNv>@yEVW6lZo=5cHjhG?tj4Rh<)uW(`Od12nR8WI;> zpBn=zsmjYCQn#n*EEVPKFML^qZ)ce@WF5`=GxE-<@F0yzKC1|;Ryynkh@1!@2VO4+ zT&OE+c~6LuMWdl&bCjKZlv`*YVdKS&^d|Y@o+Y;#KdzeqSGd;`n>{+g5DccgBYGk9 zbY=XtzaO7<)z-7h1S0*<@RMsn97N&>oOZp#^bx;8yO(`G7@>NpnO;6!U&oKM*_{W< zb??7_c}6Arx!H0c5eA<}=P9dB*gmH! zsuj}O7>mx>=_Mx}gjkI6uT^B{?@WDC<6YTH2$Q3V@%h&^pRE6DZ=)i$M!?LK%UgDF z68+ge@%hJQ!W+94%5AdGaB317o6ZD#Nlu|@QI4N#la~kn|LSSn-l0h4Q^v@;IuyEj zA>TN9VjOI?C1d+y9l1nV9IrnMGL7g8e(REM+xPspW0#VxBYcNxDwf6@{}oJ8#R?&t z&sFHz$a{RCd#7((lSEhN%_QbD7UWWLG-4hmc1H;SA>SSJ;hU+NaO|pmM9M)F9*;u< zvTXF^+T=nCgypvbPitua@GJ%&74)^&w`ifO{yeFe?qB9#+6_wIy^bEOice4KImmY!83{V z+mi=3A#x^h;C!V2ocO(!{CnKOt{#tz^H${b(?xH?tF?Xz#o3COcoB#yDy^w0y7~Cz zb@mI0?!WH`7F}IR!%I97E_M`EY80t_gh%Nyi5^tI7DfiMb+C%s(6iS-1--PtwtmmE zd%T^rLjMz{a_u~vP@5c>F0c=di&3FUk`M)lc4p2UEvRD_tjwSK$$nwGfCqxm;@ZW0;{ zTdbF%wA!rCOfh$M&TySNagg9KAHgyAZ3cj|tk&SEdIBmM{c%}e9btBCb_a4 zKLu(s>gmM=pX3^s@^jWovGyxE-VCE6(*yMNagoIah!tzwQf)w$GeeaC{ig(#sTfAz7U742t5;(nN$JaXI zhWTV=yXhQ_4RI-8=coj}^J&>6%l30J141lKm81MUOLc-J-|M1OwM=5*5D$}g8fZ}^ zU}F*eu;s8tyImdsN1;xE`=o<=4twSjr)Br|zwKf>!;6}hO$oB$j-flXOO+>owWnu0 zDC=6lNiaWWDVcIKmusjCX{3>Oq!oyD%)8b|Bc9r``OW`LqXO7PGbX6q(;BNM1WSxR z*;pk$OehQGS+B}lR4_T(!TG|v@Y2Km*3zi7#aI|kcZhB+5<(1lClC}ege*&dPOodC zn5bv%dt=YZzF^coLSpGbSSeKf4<*Pt1m+{7%4}#!Lb$87&MG+G-$CYu((1>_X_{&^ zIcPZo;H2-qj2SDyF@g1if3s=#K2`PUN^BsM}xDxgh2LCzK$ zkh020YYGHt+st)6$IHvZ5zfb&u^_ZnJ(rEI1;B4|Z?*4DeBA+9U)!$shkzgV&cayNMyBl9O>1r%*hzSl+m(t5(OnouG1JWbly z%{7gG<}wrr$p~;Zl?a(9{C_jjeP4l$TV!8N3%ekYrV@G~J?vpeOXgP z$A+3RLmh=O$e8Ix_2#!ZQmvVL8rr2A4w_mwB&|dJyZ2KY){IikX>sUvp zy~#48lHWpm6-~=G*#LNSr1C1E{Fi)rZPTXO4H{2ZWljW*K2xAwV25*51VY-XZB^GJ z_0_JIJ?fd$xIk4*83}De@@w?LN7~3BGzMSZ6cJ9ity|dF3CNkZtY*(^`UPAe?I;IH zvBDahIL8&6aPj^KA-8Az0g?m$cL`qmck1ZmI&W?M<9NgJkWf8Gg-hhEz;SS%SugFY z_wA!i^Me_kOL+aA>$f>t%L9oAzr{a`qEswc^DIs#SbUAJqI2-n@vhp%6e2t%kznOvcd>1@>-Kt~(!~=pIzn zIzne{#I;;B94@?{PX3V{0(<0Y#Z!Cl=^ShEY%@{OsxyGl$vuHz3q;O;Vy{}yQ`%`y z;tVMHN#fBDUb$Rtwvs@N7?{#HKNAx>>x-XY?km8AbJ@L5M;)~)zx_s|_@DP6HKX;d zt~ozu!OW7Ul&KV*{%t1Ts6hZKtUhMI$9Hy(=wZNpdRFz-iH2<%Hwa@ zXs7STp#k1ifP`Fv{v&A1^g7?!9vi;0){R3mWd!G++!HA$uiR!Cs@bj3Yx7ftb^{U4O_;

    hj`1#}Ky?{~+BhSO|CZnSn8dlPWGp+u5k+(}rBYgM4F;U{ zdBn?YPikI291o&FvluQ z`A;P9Hgg>0u8VjD$5-mF|ec6 zv)q5pTNioB=_zTLIvjmJP6wI5B*iz;z2y9U)`sxxN24)E1sA!O)}^%x1NoebZ^$jF zK|2%sgJDXO%&9Ir;bU??NP^{Dt!NKYo{W1iiMhHSZkMcjv#7n%Pq-vIS!8>5&;l#F zP!##|Sa47Yi4jZYFQ|OkOg;aud*@q++R2Q?hqGuk;Rg)F7bXM}B%qdf&Dwq0n+VP* zHBDgW+A}r%HTplN;;#qrobja22{lym=%g0~%llg7j2nb1n9Bc<%=JB1()QKb2M5~)6m72rcwce4e`Tl3n5^VoP>_ig2pL*zb6PfGtFO>+9*f8#uQSV39N?b<~7B~ zf4mJ>s_IR7lWc8LN?#1d{D7H?X(vc}moc9UO7t#JKrltu4B`khWKNz2( zy48Nt3>@W%mPvD}e8vhBeN_~JSaYN~zRABdLc)z0o5Up^8a9Jk)dn$Rk?z!7#Mdt| zBWH)lL_y>YEJwGk4(Yw*l#oJbrwOnD4J{XuY)aJBxcF>iVP{^-E{`q`mza4TJ{+RF zs2_{a?A{0u_BniRk$M|3sj@t~VUQIb<8(j5P4vdA9Vr=5v};cV|pg&yRrySd;~&q7HS@{SNr~icR^B2SsU5S zUb-Z$Q+^z*15MNII6E~#=!y0ao#*FIW0y$$Xx#*ha}dret?_hVwCkk*wM| zxm24k?r1>goUr88({gK)3TVE;i0fx)VX#5^sH|v=F&F)Hr_K%tqIKp^5D&1!3?xjn}Sjp=|q12^fTW-KzWeABQ zGYw4=qkAXDe!P)#+cj*$v`JDYMbbVHh71S#CINb$iO5UvGB9lDo1)6_I5A{9p_OD-`{SsQSoE_nWkEO@bPm8GGp$+J$N_G21MIWGY;^N8 zg7TpJ;8Xi4Vcb{YjGpxkTp;bT9(i9vz-Ny}NdLFqY2xS=JACB<>=WZPoY<5@ zEBmGNyx{j*68ff2AIvb*src!2+%^!?mS3&H%07EYBW;baHunPKlm6xM0cM*$WR>_@ zqeqq}_WXEALCAw#qE9pKHc$>oZP7a7a&38t<;#Qbyx5zlDDkP(64N^Bfj2P4+XLRd zAUvq>uX~2VKVclh?l9$mdGbEi{3`MIquDfTw$*1hVq=%9rD+d#;SEFieBwy6Z*G(V zDE(o-%mgjOn2L=H9}{&Skh-cQs;k%e}MKtS`*a7G#A- zo9%i}yG1exQ{4jkV-PLUe7M9HJ?E9PmU=3*^D?Vt>Ze<9eeC97^wz+r6FBB_?^e%E zF>c|T;0qa{pC^$z<#cK=DIl4aCyJ|_-@Lw|6=WlMIBK?tS?QzVFlE@Mlbu^!Q()@Rs+QImYm6UD z6ik=amgB5D$zOH(Tj_6`6f_W1Usj2kye)i&zo>*DAb_AFK(X4shc*S&B$#u$c8e~jdBK4Mu@B*7O z7Q@*_MSiw}*OnMh!0OTAmfy>51}86CvwCz*HZ=x9kcFYONX^vp?fZiwnp~)L#ZbRK zPrNg@JHe!&@D)qMh?3x*-x5bPJ@wKhg_6zVdU47lJT`0Y%fr`Ete{_u6udp!*z@^w z0*}26?uRkDyR?YIPw@Pv0PnTw513A7@{Wsi3QLB!ZUfrVxFu?{z5=p?JEUDkSwa+r z+73$X&E&LtVqxQs$^f31Sgc-;x?V05 z|Mh2|3TF;Oq`i58;J(si6qsbkjU+=H8r8!JE{i%l-Im%49U$Lbmm%Y#N7}ahM*zG? z4(txhN&Kz>Z1-|fc?vX$(zih)JUPYukN`#SG8 zTm{!iB)H8Q7k=P=ytBFfpyXUM-D>30UtL@$y1OO7@buss7Q7H5<=Q^^`+Ao4k9Ooi zAA&U5t&WSchN~sW(6w7x8axfAsNyNDzNxGS{KtG~TK?;ta?y$+yyl!;JGwC!;?v$Zs zfOwJyTzT(*=J1ERS}}qenn$RO{m*+iy@U#Y6U5!o^^ezec{~mF!C;4wOY4<-(~7^;OGn*Jm1 zezadvn4<>eJ7SSnW1h#4FoPpk3&IA*%td8vYzQaCh4+g*j!N?P%ac1Ok~<@WryJVD z3gXA>$oSU!ifl9Y>A(J{PtWx9UkqMj+&>`d@|7QaTSPgMtqrP*NaSXKBXN9hzPK_Y z|Do(>1id{=X+89PBMSi|7+a8Gj2AW;&yd8sxXAqUut}S>=*6|)q!Qw*773X|W$lk03C!im4-B#*N^P`I z)h**NRQKH_-Vb4kcm&Uk*Y9clktjnD>166g-s2%N)n>q-CzL_s zkE94H$1+M0F8s};oS66B#dBUcM__Za46+LxnB`0bURTYs3OSb*S5X=OAINA~8T)ZT1pFRqO z_@aTtHj@R9w7@$+WaYRWIqRiv%){1{XU(?} z9c)_5G`}`!r%FbTczZmEo1@ijsND|pXf@47 zf*}>tptsItG%7qdcSGK6-7mOh+;14qOna8`H3+X{1xXoc@RTJ|vC5S8O!^)(qb`CH zQ5>(MtU(6r&LCc(-(`Uc%9{CGFl+N|pJW9PZuLT3_2TJQ6O|n+le|nb&fqpA%1ZM7 z9-Sdabd{Df{g^gwSRNyZf&}V*%DqPc7Zh$LD#mn@Qu~M%82lcyC08_>TuBj#DX0%{ zPiq;j(Lh{|o-!t;C61kkVSu7b$brF_isftBySQUha|*>kS_B@+27&BO&-(`^tR!e* zC?jT%O_%kGF7L#y$vhZAGIgN8{( zHi1^AIr8G3=&Pbl`9KdxaaRdTkyZP#=pKj zoM)67FKSz(Ce&d#pSAT;$U@ZfKSdx$jbbz6*(=dx{ugilUO$SS0Q3@4@Q2_$r@NDQ z5`26>1uH)%Yxz|3WXXgddVNwnfXTS|dXoWt8yD7hQ>|3<))*3@O`H^Km}951&Dv8F z+yi+>3{21U3<=hgr_G+8&EEr_DJ3-%(m1ghu4D$n>KR}{*%>!}a-swkx(A-vNUqIa z{;AP7D)yjygf3HO4{o<+Fo+oGgmu@C%&=ym$wX7RJrK!DAnd$h#lxW zH?xV%z$6BqRW&NmF5-)8y`W*ruStP(a>m-AZraTr)-tpFH#*B*Om4!y<;lToBA%@= zb8D%RZ>(hf-`2+2pf0b+2 z3E$Lk?`Es!s~$Jnf;wa;hHq4%>$ZKqv=ZdYh@gXBp^_ZayKo3;H>X)lQlP{nRkgi8 z3VoZuS0q>w=LpsVp~5?=(L7p*iwpxP4tH7`hX%FhnJcrx+}>)?D5{tQNOUv3G{8;v z+F_svEJ3J%)%4ptdQMkcYXy6>ju4B6tHDEL$6LuBk?JVcd^g{IZgP60=wg)fNA8i3 zhBfcW=76Hrh(vZ&DG{5n?ni9mCrIOzJ+YHel$~!H1^l|A;aUqMX=iRJTGJZG<0dph zkzs1earWMNT%JL67hd95wd{4xN~l&7bYCYVCO#>&UzED@Mfvs~h4bcO&X7sqP++H| ztAcxOiRfL$HKh2IHa+lgWEtY*nkw*~>kZgkqW_VK-wzR}CnQI~vbJRvlnOPtq&riy zM!-bxz$S{j$xN!{hCBl|ggM}7F!lM>Rg)G~-H?iCd_A{<%5T=G$xpk8nvz65qM`N6 zL2f(+YnI~l{sa}ZaE%2DkN@5Yp*?CUIqp6ebdLQ*eooMbpbf}9E8B26c?Jnh2Z2D$ z?+DqAYG)mEdP`LFy;=E<1Mb;p_AwR)K~8xIYPnqJodc2M;Uu&v@)xJ9jm|nBDYQ-xdpaoJ^U&fq*hfd_J_q;WhAKoQ6-^HNi6>Ov zHd?aR3%^VYf7R8DY}YtjeB$ZCfp;Z!gv4m}a9~Gp1xeN2_Urjk33bqv5*DHc6JGZaveOTeaDYhv;Wln7=&|j`JIvm|wr$dW zOBK#xU)nZQK<+|Myud zaZ>m49&`_4-f8H$0^4OiLU>8e0bLzdlCIyaSyvN|h=SKn{y6xdFEa(iDt7NyYk_wL*|0 z`P6$WRCE|3Llm8&$8!_Eo4yGi?@;Jn+fJlcV-u-$SarywrYgz7dy>Zw!j2s}KOGei z1t(ilWNXAJ{Xi)K0pDZ7@ByL&F3-}jRoz)x;H##`&0$~M`1r{zv8$-o-Y-;^fb&(L zCZl4PQV$)NMy0FvMeTZk$0(1?(=!NA-9?5Cos~pw`M71GPDm=(C{r^6wTGVhqHn1UOK6DD)#eJ;_d(F>2xi){~8{2uX;iYOP`)bzdN0WZ_s zhL?G*1NZT7ykRFL6i4pcHr}tB;e!@Tv$#0!>z>#nvQ@~O# zl*yiaGHl%|cK=xI=`1EtWDzq@vjz!Vte99H2*uxQ@JP{gclyKMx!tsLM~UJ(CvGlb zm0O0o{-&HH73FJN(gJXuF z5kd=zW9cEzDAE8F>CA%a?{~wJGh`=30)rDOO^6n|G}F*QF^rK{Cek_qE%9H~th1qq#r(qn}+Iu5vI?yDy8ciU?~z|N?jJ}ll;tyPqqrr?P1 zVWE2voaaY;wSKwn*?vR?P^o^0GXW-hj&OKR;#r3n-6X(QPmR9IoAj_1Pkv`J!=p9C z5kULXGCQt&9}r@$+ePXbSEcj%Dy;#z7Wf`aqr_IGvi9^@rDR6t9zI}*!iL))(J@hA zL*SH+DB`9TQcVTPX9!4=Dl9^TkC}2bfXbgVNdyHB!md_kKmPdTn|6FAKOfN~dT+CFAyEcyR!H4=;M}`GY_V zm<4wk*@5W6&~L}kEG*{a=Uk16yzF8`ENV>>qE%kOP$E9vTYjGp!f?aK$GK z`}j;anx(>MGlL^Uah|tteQR)PSK~Kx6oG)ZM8UPdc>xX$T9LQm33REEQZl;n19R_R zU%2{*oB|T@e-E1*Q2RU@U?On|`{Q^HaLX@mS>_Ste=nX8-+&ti?wyu3wx>?{5XC8T z2fU_e$Yk5JtQamJ*19FIHc(}cQKn;AtjTJ>RjEF^mY9x)!$#ZC3$nc#+hFIgO_#aB zWe9x(oht4W!>)7itI$n=e}Pj(6wJnn3Xvo%apjei4b933FZ9bn4X*3q2I%V zf+8oXL#=4UmBMC6n>vZsDXq_+M6w_5s6BihoR|)7z#>QqM@c5rwMS4SNU`==xVGyho{w*t&9-ss&!RQ+pAjGWNK2ftG-E58Ci091J zIXW>c1l)w8eNx5B`;xc#awxR$=C|R$_wm-ti5J#FGx*23$g~?LES;gLcmH}TQNb!J z=YRzkQ7p=D?7u+Ecu5}DnX)34Gl*&q)Q;QKlT@?@m)p`_GH+`pfS@o=UPO8lKL~A#Y*+#ijVXi48D~EdsLggRn{y+v$gQ~BZ+V% z^)-fU*T;PjqZSsNbfaCXRlKmUW!DAUKIK3oRR0`nzMS+_IUFDh`ihXarRA@qsNq;# zX0YY_xImE;-n3S!O!Yq`6bM=f%&KRekm0^o43*3NxHOtA+0tU-#i*lKQ(t3-4%IP2 z%5?~R=q&KL@bp=Pad6rD*IC5 z%}=>WAfSMpCz9?PPxrm5^LzK_f?h!j`yoO(^Vm7er?wRe zv!-=*aTWrRs{SXX23`B2)fD)gAj;ZJWXUDiz|u3OCp1uKSG@W{uDfb9Dxu3=ioOaVe!0(5<$gRsKIyE;qa2&p*K_~^M zgS*?iXes#Llm#it()4BY&=wGEpChFFSj!XeExfwAe%Q`?GDq>-LU%KvAv3QGFWE)( z>Eu432YN=Gbp?UTwI7(|TZMK@LAI)&If9Qp!fPI4OEGm#$sLFk0nyLxyHRf1x+T0z z*~m0V_Ca=p&H~rpSLjIo3Z-k5jr+lsl_mwLMR@5;_YEv9xZa#ah^8&g=i6sz!x!5n z`)hyZP`EsLVmi%@EqrllqYy@)N~h(?bJ_43#D*)n8X3tR1P2qM^165BCF-NvXWkQ< zMa45UL9cy2|AOL2BEmHLQe^Scd0w*}-Vqz}C{Cb6H%CvF{t@>FmN)4>?kMBs0f+K2 z!^#I|G&0TH>sQ+MuCpPXc;if(W`~tGG%eqs-bwsu2%UOB@$syF`FJ+0XfV81D@a|S zrqe8_^7p(IEr`AX7k9=XpgUaR!9Ru$p-8xQal3Cu8UX^Rs3q!ojV>dyMW|G;K6Sls z_mlH^_o+z^;S|*=Yuk1}m5h~>6&jmhKj=T+Ms<%Ov*!-aLJ_|iMC+XMkBZir((EKD z;6Ji{`)GpLl_wlhz>IOp0{ME)gUSXn+cz7?HE=O2YgbL`LHHxr-wG{Ojkl;qZV$V4 z`*x=e)}*uKp$OLMca>0f1;~L zDtqp3T%vA!)aSViTTf~?{t{Cf^FPQPq+rJ+^AEh83OPoa9tkh}k+uGFi!Z4UuN#$! z$TvZ1lvE7OOwpUGtwcr{{4JNTY~uxjhHE34lq&kH_O`Vi!y{pLne}!p-V#wTC*)*T zeeQ5Zs-)AJH7<=PQ(?%|3g?u($p6fx8K(?8_mN4eXbeJnNx+@L8Y&01~)%UIA(^WLrP&t+-3qYlBnrS%F_I+EDgkS@aC< zQ+KtSoVBo|sjE;cJuCd$^Sl`xrLx7x2JB+sRs%zKA)YHZ zQSqBR*W8(Kdtlgx*#e$-!;B{SorEW3&xT<8Bkdqzm>n+OA3HGL>x;hGDK@};LyN0H z2TC7(l2XqmL8?fI>ldISPARa1{q;5;+t|z73Z6TY4rvi1+H*#!*j%o(mIu!$FKl;%Q!_o=#4YpdS_1+B`l->wA~R)%H#qC` z+bJMT4qt?B@vfEL$_*W_f~9hH7Ls&~YD3*{ybsV)ZP)h`NUVL?7Qn0t1T@QnEHdBy%I8PS8~~(95MoWMDgCs?#He1~q2y z2Fy!E+BW)>g=OChJrIJt$zKD|b1WqRcaa2*rD%m*XHZFKxki^yBBxhK^OX`_!YB0X z*WLGD9Hw)`5V}ZIwoKlcEll*uWW*pw-1E=sKzD3&EOQPqF9&o0{GiLk8#MZ{`koEK zoH&Zn(y&sh110@NfSU;L;~%9i+K&X_P%yIKovMQpU5oKYG4w2QOk=F4k=ANkPN9eH&}IdEa5R zjd~%Pw2`VhEI0u3%R?iv^qh=9{_~e0_7j=RM>H1XSeC;Oty8~_Q5iAmoeLYhh*?=B zqy-ez_kjwOm~`+#h_WR2Ekp7y5HatlezF_e(}e$zZJt392%}`2DYyUGhl{i6T$|tP z#7#g~4Nx{#-X0TP#r;Y4qxwz=@WQRJHp1H{VAs{yjWFe){=~6ThM!o!u0Cw*<`2t| z)PO*R&#fWioP{|?&9N0oD&#_i|@+JSCKRfo*7!puiv2Cr%!`1GCxjG5& z0SQ@jN=0V{nuHdyQTJce@KE!U+C79Ph|YQE~9a zig@SPWTGv5*L#%NX~s}aBnyv>TMCt+OW#>lm_ko3(NrsRbKE3W~o z+7ySYnnC8Bx?RONUx)D0OO67l<85v6nh82FoWGdL)ynIu*wDivCRSKo@=?tuIWp`vxC*!#|Oy*){h zQVj5{vZ~A&0oPYsSmbJaR1sJKjjXKy?((&jWoz~jTpl3SfF&@((kfUEIK)H{uFt{>@jBmy7O9t}}!DecQe8mA^iA7}csuPi)Zpn7Q){ zb?!5D8+fC-;C}{OpgBfR2QtO(aRtk%5cv*kUA|ua6=={b83*!#!1P2VR>8}qzb)#o z^u_B){^1L@NyMxCU*|!}mbYCIam|yW1;Msub$>Fj?%Gn=VAIDd%0^OJE4juG(W6{(b)_}Dh2tzxtogG0a5T&UUv%2wrZvm_x3;1J$)x?D-YH2lUXt)rM3577yfne)YH1A)L$ zsgm6^V8k|EHjZ+QqvD~egtbp}m`H7rZ_vbAGX|R@mh{XH<5uD?DuqxDs9|u_b|9Sy zoBJ2VXGIszlpvnr_lY4hjdc(v3`E>tF3DP&j&wS-8YBVDsUGLtGeViP^H3DWa)k@9 zY?_3iJR_#1cXa>J`bSsY9LN5s*Khd7^nA!n$gg755M;Pz1VGY2Ar?WTL7L}-Z(VSxe_WA#ai6Dz&)$8F0tWIwt-Q`Z7z8kjC8gT_9r5;WQbfrR?mT0Mx$xzH8SVKfIpan*f zlc&h0WXEV@Hz&0R57p~A2|{OU4?V!DT|q&wOk-GEzzM2tMtCj9wl{G0TIGycO!k}j zI+#=I0tHz?fgA0&dx<;MBEPYHuxAnwb|=gl6t7_&=?9lMsGSs` zfwslro7r<=q>s6?g7BUW4s*^-`{>eIS>J42m51D^*a?)_lS#&mNHCw(owpIfcX%^p zyfsuGNP|0_1W9i$|I8y{L={uaTOjjivt@o^Y;*m?iL$KHEUj zD=kQJ*hq)8NQJ5_S-o~$djN#R>yq>;B$*~dX4F7PDy*#cb;msYeQv($^vnH_JksD- z?jX-|yin&vgpIqI#Ulz1b0WMs8lk{n$~z3|Kuxc#MPf{Rm$_V4Hml10;Eqv#sQ*eq zk9uLD1RtYw&1?xGcqu8DV!hf_A;3#`78Zed2QT8HvuSVj_HmiI?QJ!f%vgK^BN8db z605Rr_Hx;QJe3hc0Fj=tpU<>BTzrQwk5|+Yy>%St;>w1E`1f`Ac=@AnVX!(ZQCn8>=>mU@~n= zr0ERPX_Wb9R!vI#)yV6|^P90Vjq6m+AoN#>PbAL(IRwt;+2Qpi|GcC-)*ByQQ+$Qa zvD@Lw`4e&$(aq2>?=?KKi955tte0CbMg3t0&ZZgsw;GOtceBl#`v}JWuV_3de1P!K zMdIiv1(N^+3H*kOKxEy`Ol%pjejq>Awav=5DfKs48Bmg>fJYsPUAyodj&{W$>@Y`s z+)IF5Z0}_mclIpk6=s<>EIw z^3jwMWeMxZYp=`20vK?h;uD$Pw3feQI{F3k10isJ6*v=w)=;%NaIjPzSxDT%h+}KD z1*ef&!5(9Iw_QT$D(!~edg#%ZxAe*d_6HPpt}z}e`md)MzmSj^+HSVpyZUQK z5&a(St!y?wI4D#c5(4fCl<$2@{6t)6NbFI*hNK;|{tokowhg`$#JPc%$d!lo#ne}@ zo)HB;M~U?Xfi*HXp2qfwlipf%yG>(6Exe-V_ zBKN>r(2?PFJFy6YJ7DK&`WvtN=-@SVL+$?``Y5KO+Ft`hJ9moq-fWX$l8uU(OtYG~ z{XTwle|iz`_mR{5v4&Hf-q`1*viFNM(_M58f5n>Tx5e3NKT{}AFhP}JIXnuWEib9$ z)?p{LF%l18-Hvj=il-8c(C`r&I+7*O@Eg!G% zHj<)8-fqO+I{a!MjxS+%c$#J>F+w(zEXvd80>^nf2IdRU?REPQL`$W@Gk!!@%sad5 zyK_1CBaL_xbPpm%C^SZ9bS%nDa+~R*?U9N(FNFQuo`bd0wnRNURSu< zYi)o{p>5Yjnh3YxF|+!;&6Q@K`(9}H&qYu$*p- z!sLdd2Huyp`p4|crf@&jhqYErJi2;;HmDX~bwvCcFPagPdzM!|uY%4cCu>~rX$&OD zUbT|A@p$-1jz_$*huHnW78UoGyxi`G19#@_%f<&!{zR^AdiynOg+=`(d|v@$7~hhg z!HtF}aO=%tx$|~JLqSZdND(4qCn_=Hm5NL_D^EWh91itv5)ae!ko1%ppzpQtIOpckmI{067I80&r*{?@eK|Ur&uXfp)}~8SU`7Tk^~?RM6U&f z^iVsl==ZNeUau@2+I9vvZxr+++h-GvQB4gR_+oEeq1Z}1H3~9M>+LhBGn-6B@_i)Z zG|L=zypG_`nr|yb%ZduVf9#Yh26}B zX1Nxa-MpmxRnviM7Y|`K+~Bx2*$nN%Z!&%+{IRvqI->5{pNY@8f0&`phHvD@B3{!M zG9>vmwR;RMA z>IkGc?X}V_*{K7;t0}P>D&$p9k>t`RmXW8*A)y-!7y77Dy!E-)0)r;B7rDXc3RN9K zF&ae^gMn!xm2ktgIHs!?3B_HhqHGH90f2k2rF?Lutj+wwnsor)UcsXkbbuF@cz!_2 zur|N$x3+&LZRc&t{oqyo?fO=b>Ic-HKrUxafp!phd>cDym9egbZ+RFf5(RFmxSq20 zx}YwbWuy`)j4{>-T5tnn;H}W64pC$7s0-`rkG4Z|16-Q$7W+@dIAW);&?d}eEXTG~ zRDR8q*_rBp}C-eos|V^ zVd%KX(_SClE-iWNOrB%8+mgty)sp3pM$OCZ4k0~IVKVniO87}vXIoELg&-t&|G)4N zfGsbhYtejr_nsamG3yoKx{}=)B^^?@ULT!panL%5QHQ(c$TO8jGY7FfsAsS{@lz1G zr1Yk78kMQFk6Te|Ilg~EndY#nr5-UHQ-M-RB5+oK#AnlDn=pQC>w_oN&ZHa@p=Gwo zd5kT=S)$~`bj6UY+iK2a+8e@S8jP@I4qPqmd>zl6Pl{EU5L8zoDfS#oZ&CkCz>9=eI-|*$l zu-tSrYq8>8A?qY9z!QnyFk}G$1H4R!+fux=Bs_3S3Kp8X(BXW2i6`z`)wUQy3agCp z;GtXSe;yDsk+bRj3(?gm-dtOq!r z7qx;{L;klpv@^Zv2r5WC1c@}0fWL27C@;9dwGwvhUHb8c^9DaI)jGT%couUAMowvm zT}2Ka&)=M!u7+Baq6hNN@W>{;cVMn0<6Q2Q93boXZp_B%p1|@5B~eY1#D+OkET^z3 z3ixv8dfs_Hcc&P9Ns<<=hph7t6AcNvPna9HwmZGO+x|76VqFhlWOcs*j;fVk5;ae2 zzNH6I=R9J?QV#@&Fp<35YG&#%V&V;zG{o%oX#^ScW1lA~6H%3k4!qE+1X=SRr`ANv zJI+`9uN-GE`w}r#mKc^6=voWk&8O!?yh#iDB?AE4EU>D<-}XZ&qNi2QVK}luF9lsl zK+R=;+h~m^m9(+X7>yMsPYBSR3t5p($0v*VV-cg%Ts!2*uu`67J{i;=AxE677pw>p z{u}5ICUOf9ZdI+6vtgJ=>fAwKr2lnka<=6ao@nkW$LsAQfO^%9nG!Cm+DSPcOW1VH zg@1Q2ZLcv@{*%(x^fo&x#l{yOS$|no?Y|bfZN{hUi;lO}d+T?h2j>YI*C6C*&oE z_WGEQ$4PRkbp)dc=y`z z?_3{Tj+6^ji;T+TZ}pd!$*paflH5}X`KH(~JJS9eg3t+5u~wWvx90Nc?`x$CHpaea zznTTQI+@i#-iBDhf`de2sBS5UbgAfU^M;OtJ8cuzZ~i>qp|FVMXyJXG(6E0P^@ZzO zaKlW;J$f?g9a8;;IILzM60`*tQ0WS1wecG)5fK2vr3o@+Ysn|~(myXMR^MJ0fd(Rd z2UzfrI42}Y1A8WCQFK~jl*bk^(ET^S-!)ocNL#lY(ZSn|^FDeAk32lH&jGfp1<;EP z$Ekv~C2>?GLO#HcASm;yxuvL@AFF7KH1oInb+%v0xT*$GQulnn@#>#WedT}hi2LYn zIlm@sxEu0U1M{;i0y)}@>FRRujMjoo;KTSPo21Zw>5 zqkK^{Pm6lkPhcW`EB5K9kr|d0LO{M|G|SWFW_zu6KHst5c12B?%p9irN=z;)WAaB{ zFvH0uZ_j={NaOdKK|wBztW>?@*?%snfMK@nX!zBlpZqm~9r88cVb0*Qc4KFL_fnUi z(L5=&DU|kkc*;d16X?>R;L2h^stayRj|IpUE3??Rv-o5ip zRCuO}haE5B-VZ-_tw@rRgRLW93UNEew4&I@#-R(fqA#HY7$lY})A;YoWshWTgY2Zj z?G#h2iQ~uW@r}VEH;n4b=uC|Qgc;WhiHc6rA0@UuP5oIJyN|%RVWtIZLINC2`y`Zi zn!NTCP$1${khT&p9H-~T5RyRlASl(e*6fFeT#uq5W?b@o>uTui5ql_+44(|`Ov@2J zKfiTEa(e&5+cM_h;W~)I`yu$~-Qgd1=|db~!O7P3#4t}A(xejK?gd=p+{Vr26zY9= zxdIY%+Owzb?xPJ2UzJ}}W)Q`wkN14Lvs6f2^Rsvnfpjei#>Sf3_ctvVkC6ghbpwdn zUyfkrD%oM2l378&XR_1F0Jo2r=y4E6moTv;vnN8j>Vu=YP~1_UZg9b*kM#5v&P}b` zFCJMnAcCOTQ+zhSA#&)a1uI>z1BEX`MeRznpoSGL?30&hY28j)uFV=0Zy~>&ClwU@ zELiRf6Y81$CcJ{LDAz;o!DAgre{49hF+cIK><8p`rZhVddaVzj@7+EVxR&|Al~)XX zK#dH+-4UXFDXiY8obAhmj(rJu%D*WV;k^=VWU0vlmc~6g43c zeUD>RdKT4doPv;%To+_JJhgVj!Ld>%=KnB7o`3SyN-b)!_b`vt6pawXtNdFxs&79-E6N?p{{HCc;9+LPPOhl}r5ke?Dz zF}U~wC5B-Ce9U8Oz2li8OvuRH2^O)yojU_)1kxpRUuv;!dC}cb#^YeLf$W_YsC-N% z1nnV#s#=6!swJe`i*YC!9S>=Gn^zm>=j=p=;^~62kN&694x8r(&wkGak!k9XFP2jJ zS)i%?dn!IBgFdeNLO;K#lwxPRNA;%S%!vl(5@S75rWQKm3cDx-8eCW&vN~IKP|*`c z+(^Vm6jqLlknUg}g&RY3LwF&^J0_D+<5bWAx|~LS6kb4xAm*V~nb`d69i+;3Yd4o< zO?8;7-qX)v%I+>ilu&(u^d}8p`S+%&f}%kes~+*XpYLHcn~N>MW+qIYLdw;p2~K)w z!7!ptOL-yRbQ(27YuMSZhJJ}+Jmub}o0XXDu1(lQXp@W7xhG9#k-TIwW-fop0~^WX z8h+tEbBsq5AIuVPSA+acP+At5)Vk&y##*FA%?fq`U*#*@toC}D6XmAZI|2%z~= zgSuQhZy+=5;?d8??pj0Xs|lVhgd!5n1Y{$g_*7m<@?*8|%e`4I0~Ik`rw&<85Sc?% zAR1rY%*P=)Y9hbr{&roTrLwTNWE4*o`TS*GS3;2*ds5&j2~=O*3DPtRXkA*|l*X$? z=Lx6lkkxJ|gDJSSm690RjzYVwL?-xl$N?YVbGFNJ6~x2F@gYzL6&8|G7KPSSYs|*pvGx%(s2BZz8w#A^*0i`nkEHw5iuj{Dm6pk zGkF;7sH$pyb||A0+`-u~kf4i+D5+4)!rj_APssOyt;BFjV`uD{FsR(_`Fyc3>oR^Q za#7uZyX65_)r0Etq23SkKK9uf3dohIZ47?9rfVRSa4ybzjR$brOVdQNLrCPCljOk# z2PIaGHVWxE$B;{#>zd>fP>^`4S9Y}#wih+|Wf9fv95@4RZ4RRR;g0U}D~PZVF=%i% zaAJb5b4)v;JxDsCQDu?~&pDN6;6l`1kqUA_b0O-lgRS!^H+bpimY_zK*1UJ7!e!(j za{0b9I*w1JY+|aC1g-$4?7qNExxm8AF9p!$dkp(AZhwqU8r}A99X>dD{0{vY5`|s>ho?CTW4v4iOpeNgA_VT zyGu?NQ1DpGt}J$@Oj4nXeJL)15T$ZIo-p?y`_g6E?zGYB+o-9#xjvBdOiSP|Vx^v6 zgH={$qnDSf&=Y;)j$1-}?Hh9_3+At;i7m&nybbb>mmnT8BS${?V&$f{xt|$M4-p$V z-Ur>cF1g65tWruq8-B6s18BYJa ztxXnaG0$?o#`#H^YW&Tdxd~tOv@$r~+&9o$3DPsr%b~*4O`tq@ANTB4i&3o)jIQhX z?a&Q#*#)KQDx(E=JI*~Rzt8YvvTyD8HNade|03dc6}{~FDSjW?(G(`b1M@*`scYt` z1m#uVko3%1PnUI^AZGRuwsHR`==G65^u&kT)R|Q~APn2q=G%s?!+Rgl3c9~&rM;EY zqSgnq^v_C_t7PLedB~b0Wy^&jThNM2*HSd-f@H!?T0m13C_x%U{NCAsuke<0?I$}I z;J8qedalfRJ~!1xzsk#K)|dh05L6V2S^;kf`i@lGQIEsyyq~!O^cu9@K1gD8uX4T; zi1u?VKdjHf4ut0XJ9y5eqenS|xz}8^@b>Pat-lFVBBHT{N*Q6a;gspocCGd2>8l~S zz_7y%JA!pcW9eJ}D5uK`FvyY-+_5>+uTH_owNGN785}dZc>$-Syts8osOV)$l`G}7ZU z#O@7;PpCKwpiqX%5y}|D*QjBX>d2rts_=IDrCg;SY}L2aO-42g%rG6@s)NJG$&YIWmpUBD@I*&SzaAfHF#kk_x0v;|Ul}4?AHS`=5j-{KWaL1v4Ws@`yZD!}#c8^VWWQ0*}QTy><0gb}ql?F|&UaRKA zS3P;k2Lin>Q&ljn7T?SSbG%I#rK@t5Ji-Cf(DEKzT76?}(TDGIQxpgsbBn5-1Rnnq zB2m%PKYjRS#h(DBS7q*RtxYzEh9EU|`&{_>2TBhkynm%*zKY1Cs>8oRAe%+o1f(GS zv|TSa+8yZmy;eL9h#XfKr5_%Wr-}iG8#PRf4#@h{dS_I$WnKU|K*qoQ9`1BsWaB9` zb)dfU?ulK-sGS%O%;S8V98rIEYVGi5r8PCWAF(1t<)$lOT;?q%yZ?WM${#xtx*+@N zcS>%3$b-{y(5t9Cd}BoF=2A$L$8@JOHujIP?8Q(7HcXJaGmCF_s(9pN@{Z6LYWBx? z-^?~ilUysy8B}a`7mQmsg{S0d(vmMcc7`n2neB8Q)8`G{Fk4_CdRW1kP>u7d+~^MP zSdqAlyv#>}jX2%=j{`xjc%HWXbTZaZiT#1&0Mi4&9=#t{!qWogLb&bRSoZvcY| z&z|}%)V?a73Uo6ENr?Ep%!E%RB31*W+fAT3I>&*D9k-vb2vg)b|8Q>(71(G<^F&Ou zaoJD-ew;{ZX^Ev}f3MEjSYnKt>1UP*sFoC^&+UZIl13LyU?SX~Fh3AY#zEUfXuJ`eHYmG3knuulHgjxWlgUDLP zlMD}-07>*&ErgFmvbB;L=quS*reiaTpEh z#9iF0jxd9%C~t{!SJxMXDUBmdtdZk))|Za=*3{}QPE2li1(5?0kI&9H^)$0UySCje zQ3Js)ae=SRLZ+h_-c&cbg{-&!K+++kDKu{+9qV(!WzO=CsA-ZZ7CET?i(0*@XeDbS zwFfhqk{?x@tQTeMXaxqQVY|#&luYB(Uh@TvG3uYjIfEUTUTqQo!R(ckpBGN6r@4F2 zbj!S3_fHrfWO2JGs2hqIt!^cd{?L{!W#5$h<^5U}W&TJErN@1Z?+VN2c@}@Q_^{pr z0XzEO4Y)c1s4Zt;rI@zE=h^Sfo~RL43nO)7k6}(A{c6lk6|M`(6>hq8aij5Wf4P}y zBr<&dGw_gjQ#FCICs`MhedcCwLdWi{y2%iSt(*+&U3i6%!T1rE>EnOvqgBh}?<==m zaes%(7f_WKR2N$H>w(|lsJr{I3Kl>Qz%2kPcNN(Tl~o7BXu7ah%PJ*8bi4 zY)eM+8b0Yia#9615p`B-zA9}y?5UdW*msE@l4p|M3}ftye{~XkEnK&lI87zWoNtjR zAodB)=uqk-u8>E_8}<=CO)aHW3$u%2-qTE_#eA!C=5x)1pjz`P2|?8>r%w zD5`h|F}|}~sNq=6i~wrhK(Jy}mu=N3a#lydV==7~`pUc)B>R;1#)?|Mn$>TeJ*Q4N zmf*<^jvGb}G5CVTS~8seqqkuWwcI)DKLkP%L_PuO4zB4#ZAaWF`qP{Uyyy`sA&-Lg*1m$&R5b`Fsk|^p`?ui?XRCFA5>`k6(NhrXDHw%ndmnE9Bn z=k@XrWn}*LFT4Oj@b%*8Q|at76koGxQcuE$BAJOG-(!$8b?Y(#8N^p2#q?F{+ILqx zICOIwvXJ>D_|^>+&R7Mlz&csr@ZPAc$-B>!p#uXOlt3A}M(K_4iv-xV&7f&2yb9K- z$lp0|jWdwbgu}YPD`T0Iy2?;y?FM;G_L{ z=iRf!cA&W#Za5*FkK|4OpGqiPl{X!C&gd2Yjm$`Gg1!vlvi4%rP^A3`9M)uRN+WC@9xU8q{Mq4d6weNt&1Tiwwbq)Rw| z=HUgRnOxJ;b7!n2U{jh?!2;w#iM>u8X7~{Cm)YDY>rGt+tVt~fKiFoB!6L2AKq^t{GxQ{rbyL^xdFJLuHQNYqE@X`c=5TgndB)S*?dqbqc%Jk z!G|kxsXAoKGmSV&`rL&qnYOmON5!+5pcpxg0Ga4qLOaBKfkkhubV+mS}vDqi*xH zRAd3b9Wr1AML6E3;&|ApbDK*@^I;T+eqRt!N+$@y*{8P=)40(3yd8DMXAOIp_*k3tg;7+n8l2G={P(>& zNQ65UoE0x6mK=Ywa<+lIQ-eKJ!xWw`)Y-XFi?%#sPQ_aIgFSu3vthTr^%4Y241*4SOV@SomBX@4nHo7#ygq@*Ic>t>{5_#8Aq8 z*_RNZ^zJU4TyA7_qD*tX_a?~}x?PI6yc+6Kud>}wGO-9~(Ys)-=a&gR0wTw-;vt%A z)!V-ee#9Ty?6b9+k7&34B-5mDeX4TR5U?ktIBS%qEPn3Qjp|wpyx(6Zh_5$%fju!P zuaG;DF2`Z|HTeleeJ+cB1;Sa2Xpi?PE^LQlGpGUgTcv3-=TN8-32lj5LcxxAEi+ZD z4`QQd6Q3|k=J;`QR z4PEpHa*7S5UN8p$7h_H44N)Whv(*N>fL|SxQiVStgSbddw&_p>tv;`y7##a8b6xiT zB8Pubfccosy!oit{&8;SF<}ZY6V337wb&QxW`;`HKBUETj95xsfcOQSgxO)z1^hhh zs{YEzHGW%D!eG5E5S6(>g007Ag>KC}>Xx)V^ z$-i)IOf0qJJAPU!b~5vM6!mM*B=<&C5rG!>gbPuu-74bDu08n9keOvFhZEB=D{+^e zVC$X-!lFojN!GCw0vAX%fC_2RgmLQFGicxSouJl5#WdU+cnitk8`Qb9LDbWHKTtU( zGIm<^EU<3}3?*T#1@@PAi-7rdK=HnkPdS%9NKbqMs*{i|BwycFE-@7wg-vj^FWYa# zQIBVj(n&|v3j7{7dO%#1iy;|LBlsobAjvtJ*mI2b(xV=9 zqn3AO?qV2EF(?NTa!Bf1sq}!f(P~N6NdK9a_i6c55+GL@%o-%7#vsGj_+P!cX?L%7 z0uw4~mwL`a^V*NP@~s}7bF9})-iw*#AN%c}?>c5C=~Rt6j~gQX+%c;_c$Kj!_=TPX zWT%n$99S?=x;Bzk=V-U=+v~Yd6%Z=Go~Qq-RN)$m>54;L_y!mx!*{p}irJ zO!<-M{sOz!;D|3|1|3K6W2Cqd_X`z1xB!EAlE3ac^cm|9M=?40BeBjEml*tPWY7$ea+tGL)5V zbdBcDlp-(xw*j{(301CoG6XSzeIDT$3>1(D-~F|)o8s0!^9N{?U6G*{>B;(*ShBzu z5@y|smA^RmXNA6-xWI0+U}Y|i|lYxIN>c}#tpa;KDtt1)Mk4nIna zq4Q04-vzrRj8KjiUS;9j69`q;%(Rm7?oncixXdONN#hJ>LCWEWDl2!bfsD zzB^1`S_lGbcV5bJXUCqbEAS%hbh_tB<;d` z-@wO%o!V$swrE$xH+t=o125*^e=S5sD!-QIi#bnAaeZsjW(5gUyD4qo2VuG+pJ#Cj zp+7CJ!5A3^ZKaVuNPkHl6Uch(*Px^;Pu{P|nD(_D4%#EX*O;6{;zz9Qs>swLxH~oU1!K6vVF1vl!_jB6^U5Ui-Syk4faB* zMQ#U`Z{StEEz9%=SwPP8*2|@(<^VT?M5P$0w+l6)JhZP#|FVOe zJ(!JhF#g?@7+f%F8pYq`J0u5iR4XI#6<=HVdm*rsBG%Pt5pbl(GXGIA3_@gb6yym< zj=@2DHQLtx=jJ*eHtA@IFH)=c%F-0=^STh!ZoyBKUf~#6m9wbx0Fx{TVr}870nK5> z`03fENXqERIS#VvZB;>+5e=WK6aY)WGM2|bG)nCDAn8a zB-;YeSF3P2UcVeb;#9qFyrBHMVeX2s`)j&MxgMetzVOA1)!@Nr%td!zR$>cZ@q;XP z`%K}1*i2^~aTYt*5%#jvGW|)(b+x?4@Vm02E$zB{9zA0-`zCBg%QVZ9O=y3bajek0 zQFnZw%^lyZ}aNr z^D)bb7VOqY862~^vq1c{MyJgw9FR7lpGRQPqm%?Q=}%l;aA)2VVR=c%D_P1?{k2Yh zw}>G<8ou*0c&33>T9|autU#8dj2^MXCQS_6GO583bI*xDX7Z-%ADGK1YC!TX8OmZh zD&s`VLZkjaQ^Sk!v%(IKmJ&jg*`f_XR&6hF$c4an0}ozAp(0oAO{U1QF_c5<2^eL) z6I4@f$eT;U;eqXb*F~BVcjDhwSQU)_IU4BKNNX}z5vM`M2z~w|eV z7geEh45!yq*f#>f`<1i3;9>JSmUA3RHxdHTD_tulGibf^Ff2;yXJ=)(#+GKeem*8c z{~<~;>ALtIw8^nNDk^o;+fo>L1dhGRs6~t78(UC}?4?{q?^H*uu%SZ=)a`OrJC#YfBJPU_o0%0$uKTSJv zbp1rmN~c9Ayu_kbmU#yI7Tf<+vj?lY9JwT*7*p)%$xhcHXNdWyd<={?@T1iJlLni{ zAB)&(I0GwCXC9+*?t))CJh*;~AWGyplWh_3`T}Y@$)FvIf?|pO{0!_{O$}Ssait|aUB@mCR&9)w7|Ak-cgF~K8dF=ny~s^Ju&`VqSgyImA>z?CjAp@tWNNXre(w}AH^b)QTp zH+yrjNfZ%JkF;Xy5j`aD)==E@3gckI@{*M@^U#To{lAYO&b>cuE#7}h4i)6H@r-5d z@GVE2K5%`53NO}9%4Lc$!dR)PxUowdgWk>kM z^>0aj!~>@>H~XFP#T{cSdvE4J$t6IH#~2B{IGXm27InLB^ab>%e)sTdQvI-jQZ~uT zh8)KR3Ve0}S?fZI+frAzn0vOxqLPi$-n4KLSKspDd(wp~tY&$9++2uJp5mN< zq?yasM1`gKaHSvICYWQeHJhjtc1TY80d_~vb8GFFGBg@FJlfljilFP%yaxpBuV_Db zo6Yu-AR4IFGTY#2l^SKG3vx+f+Gq0u@^W~`ERK06LmR5#@@&s7B~9)Ia_0k4{OG%~IhPUTOOtw#vql#v@*)-KT>y^LJw3PR72 zFHn-3{pDmqwffJ~RYps@5DgrJyzVlC(>>kt#$j6>wo<((0eL6`wyRT^tt>}sn9%7~ zFDadzP#a;@UXNQ}h(AHGC*Aj%rL~=`kOTuzI1=x?kDd6ha02WdL=Dru{4R9{lQQ&m zzIFf{bijy{D}(IaLze*uBQypHKfEU(lt1c=u+N%MH~l?N+Kf6D&mOMwG((`QS8S?i z6oF7qTVc~~MNr*214e(byb{RbTUrbf1m zq{aw&QP-F89jBLd0I#Bb{)O!ef^X0E;ftNY2HMYF`!l3s5BjA*5lMB}Fk#?U4`d|C0QJWv*C@zHVW!IbS}ga~l@A&{lHDU2wTXW1m)AG8hpc-N z_WrbQ*tpHHFBp+&a2!8%K}Jo1=`BJnwkcnDD^=QuJsQpsl3D%Z51a~Qc{aR7e%d@> z`>H9HFUtogu7G+gRe_cCS1wXV)li`w>V`R^Yd{BM(Ga!Nwf<@ysMKypjh~+1EBTEW z(g|yslL+>56+x(oc*aBpVE{2hC9AaFjA@`h1kOjDuz&sxZ1n^lgwy0%5B$sfC9bI) zJb-7Bry`98m+fm&pW5%6r-~f>M@8s%qCn$#{abdYeD9@t%WE-v`r99nf~%aEkTX%U zP>28u;EUws;*eQ7uXiz3?v&Axc`8QpLasI=>(pmLZx7F}iHe-8tY2lemyTHT&{)>fw|;wZ*4=tSV|kC>+lc)`CK)n`=jL`1K5t%wJ!abZ z-{NEK{@RAGqV)vyO*-HqJD_Ayntl#DH8Ax{&&U8`(I0yCp9?rA0N_7RT5KCvVzfqS zKECr<7vqDF5e4?uc4}#L#=*WZw!DvaOu;QV_$uA=iZ9Yh>sz8M2rWp z|B4;^`3&G*Y<~G5^jVnryX~hKwMfUT%h0#H2<<51J?e*=$pgB4L8?4e(}n?;&LWdx zkd9pIjm;9>-m}lOh*8y(CB7f2a5*Vz6xkp=VwUbHA*GR8WVRu`0y8_KYqQOozOUxI zIl{Lv;Ni@2!l9Dxm;;I}c<@zm3Ykg9TT#?Xa$5x_7R#VBv*!GGPJ0=x;F}N zAE%H6FxOE)45ojp4PKO7kOp|jL|Bw=8ltPs{hc{$Lg~6tI^8#Nx+X{?XQq3nb?`0k ziH~!3-7++M&uZ|Sql12YV~AyRRr+=Epx_c8JoQ>kk-7n3j#aQ9hTv%CkhLtk;bm|5 zipeMOh&&WX`{&>F6N%w;4a54bX$?fTWjYY_-X7Ag+JhOuePp&^hMYz}B?7IG97jC6 z=*kVSaUFhvkB$y-&WY_4EWtmz0b6c6Z|;r5D671(e%Y0&C)U&o&AYo3nKBBfIL)@w z2{<*3p=jbx1~L^ou$Cr9Or#t#@Qh=S-dLp0OlpR_`*DBn(9~u#$S5AgacgqA*Z^+} zSkunJrKE-MI)SQ*fxf7&CBm_8kW~T)Ec`Ign{_|4`(;0vs<(F+t=|Rrb#ftv&Z{FJ zqh?{W@nYc6gJM|mpsT^!=uw`&Hd+nUH9aUz#$ zH@DXl2PoXQJf_M-KSH|6y%U<&UHbQnxCS1AP=qzv9;?)wK>-Y{<5cOuBWN;iti1t= zyS5sTXJ}j45u=oa#EutZh?UGTy*CKQ!<*ghqzv-g$1K|fwvd_>Glh>@haJCLIa&!2 z9!9rY>fU#-FCTeU&@FD06qYlJ;<9i49*Y;TqsbcuBW13K#oF#4!Cr6T%d`*zxpSu3Wl5Ti^_$E&~s);S{ld=_Hiti%^O>;B$gg) zy~TN!tX}OFQnE95Z~K{;E#L&M>64NTOhwP-P`g0jEhEk`UirkF53w+4K3!dNDMjXz zWLobAHl}w))|*VtwLA!Z_bwr293Uw^L`spmqicf6p~n>_g~AgMIXQ zTe#k(Jv`!{p@k?<-Seu2NX)XJJdp_M>Fs+ZTzBBsn4x75@^>6#wNMhmKr>UQX74&Q zb+oJom#%JRG%z#{C=Y;pkEWo-WYSHVgh8>?JFCmloyqV8UNNKHiPDgwcEb?rY$wr+ z{C%B7tk+=k?2fu{f#0UMuxZn*QZ9>yr5ZBJ2|*=k>GQGs-{Zw++*N*}F8hCjxY+kg zk-zJJ3(`S$v{TW2_9mcGU9T7TCnw9GvAcPVV>a%|PLA>le$AAbO|(LT{6)~+e?sVG zt&vFxcA=DlbeCl`Dt*FOQ+h!wWTA%n$E3KP4>2xs!cT<3Zap&;l3YMzPPsTFiz%kf zp7iQ1WSlN^?6eFKYD2B)j=q8RrNYf4HI7?%G4YwZqIWSuvmc2L#OhED2Nooq9jp0j zMkKWyQBsY0*@xoRrl(5|yU&gRGtB>eIyeAp0eE-^4ZUl{pow~+ z!+AI6N%I$Qb(l7>L8y5v%&ii`ABV66;DcCDk{#5@q`EZ-PDH`FM6L`w ztq38+ih%#71h6mHs_v#l95*TgEC?0}I8-o{P~BZdPdB6AcpurV*_5~VC8dTX~iJ+|I1H2Qa)dvnVlRc<+Gs&1JF zJ_xV*il7Q@;;lM5VOmYcBBQ5aycCivVKC5Bno>qCc+@>Io#hwrtW&M=6QocBHuwn~ z4-CFvnn2~dboK!q83otCc?${w`elg*&ZJXqTS~sCOl)Z=daJrl!dXli%hBEHL0qX2 zYHuYJvpCnz?%L!-(`^`xl!CPJ+||v+ojkXNIW1b$%51bIi{}f$2ni?tFpue^t#0`0Yo? zI6cZjrvZE$G<4;;Gzy@BhI;Y*-ty1~J^+X(E`V`Wwf{tS4o+cLS|l~< zsfThr!7*in&?hlggbEa92l(OC1S_g_dw;h=FNglZb1S|n0EXS6Rowd~9$0{@Wk-z3 zRo-Gt5DqJQ9g4-AW|SSSsqEf5)rV`i29=MODb00<&)jv*l8z%Bcz%6xGlw$4=(jODVZZ5r`>}VnTbMTWDr1!xQB>?jN?OHNW_fKWJohaHclpv!kyxKUn&& z;4Ot2SM~XH01ggG?z1rAac#+Hy)RUCeFyq9`BM6)jy#-H-YMRnh_J%qbyXcr+Pyr? zBt#yh;HbR?$1aJH9yPA?K^Hm_>G}R3vqJ4HIve0r2k#+~SZs3LUxh4|0-FG3RnALL*p#E!!`umB)>01pJ#p!XG?5X8)qFE?zc@1{e6Cz5( zS|KQ#uWwIOgc~~BrilizMUM^ZCD>{vn>iF~-}8v?8$cxz(i5dco!=pgBiX_5uj*bd zyfMPc4sSOOsfPn9r*Rs0#qzgiG8QF66KUobuwczxfzLUHJ17f@nmG@)qvXiipcle% z7P@(j4s!(wMIOf^m4PtWo+@KT|H?d9LXQS8eGTUTH4NzWAE>t&P(es=Qv!mu3#`z; zNxpvHrlZ2cjwR%0rKk|s21oWuWRvI@=Z%Z3VJr<#y$w^idgMd)EX6Q&;4>1OFf zfUOPebeK))X(4wJL+9;_*366 z$8Ozz<8q&{XfL*g%}IoECa<_c*Pe@?@|PC??$A~2&k%uUy3;1-Iu3ss!ui{81|abI zUN61%`0qDO$TFsnu4x5>We{bWAC%;Mlk`tC>W^Fsz3Rv8&I4J)t$=Or)mym7e4M4N zGSY#gx|4ZreyU)|5AK($+(wIG>@G31IcFnFft59ETzlKu1v^etS?69I*@5`WWMKPH zx3Yl@(I+idF*Dh&TQZq~Z-9XbgA6Tc?Ho_gR-lnt=Fw{{+qxPv^9kPLlu zNocDl5C7LfsNHFwgC$JWbTFq&mKx=)zxtDE=saic%Pij)*+Zs774PBJn^tv#EJJVF ziw#7zdtP;Cuzkwku*j+fLTf!$3{O{1#~f)Pp56B>s+Gxx*Ysp==w__h!D+lkb4@GU z1zPI8Ne7w-uZ*~?3;6X^F=xh-UE6}KNKqBrnr(B__e5cejSqXm%;TF7eJ$VD*4=kw ztig~Pb-Q!DX3&>b-)1N}>C)#6Lr4MRLMaA{sH7i zCvl^$AM=x$#(^GUBJ4H6%@g$ayE+ys0O?I))qWwK`p8Pw<4u442F(31>Zo7WQ1^3fJc?rljlk zW5#XQ#^7@bHlrR8DwgySHWJSn6$j6KHKqgmzC3;z~Ya{eRV{6t8eXJ3tq*2yX& zm~;7tvpc_^)H>6r0Xc_!D%+0vZu5oq;x~)*%BpDPRgNtBCsyU5oewf@_87t;(fePa zk@eqRQmc*NQ?0b2cD{z#^<2`GXqi?wH`~d%W!`C{1@*pVwr77Fyb4C6#HzTwJiCG zm&x0ew(nN2=-$=JyFIs^ciFA&Gde_qe6GfSJYs;p+^8LUz2?dk;K1w^KWa`to7mi} zcGp04GM9?1vLgT+9S_u~o&D+wV)t7o@P3H}NG)C;?&xg$d{TZR_JuP+J$ovZKz#Zf z%Nj+Y&Wi#IGOcp9@uQjnaZHH2v|R+?qPySK0t;SfN>?ZY7^!FWp=%xA_9Y4vEOGfV za|excp=-{*!-rJ!HtOqHE6fM^D3z}KN^ZLwAV~|}zf>#m9Fm+$G#UU4oaO7(@uRMy zyNk}5zGdwy=Mu@Y%C$~0F-NA{A9^VF^_u(HGX~5Dv^ThaQ-|rH7^iA1fHMcxeSBkYG&<(RCsL|pHP(UMWY7wEW2xXQ`=!N zP@9{6dc$OjB|u*$=3ohjtH1-Z2>;bSwqiP+&+1NNoJu_G6qo4}EHH_}g@+Q^-#3M~ zFJCBSw3S7_OD|i9ECAyUR6?qc)fnh_9hq#m8ogkg^wN?H=79pT^F_cQ^}lLfD~UO1#QRA`MME{T{1`o|3vP%als_j{;pZ&b-mG&| zz4b*25%I)7O>&M0^U*7@V&6wote+?+rWJacH4PSw^KLmmTyQ@y{$;HL)HHSih##c& z()n?u*`Mx8YM+(oQrYlA2JRou&(fCm)-@5|avT9l?*yi*pd31rlcqVAMKekLObuNB zJPGB0{i1X_ zU_-4q21+57B1q!f+0BO+0@`%h^fhA&#Wn-qeSJ}MCInT_d@H@H7jIX3`wD|e%2;}j zPU&q##1!Hsbk2S%7o~g5GkZBz!0N;+w+zOQsqgu7)h&*8N9>}wnGNejCO}r^>>4*^ za4!F*-`nwEYMPQ7WE-6hJ@_?l14u;VlAyc)ti3^#MSi}3c*IDr7g6VZDB)UG{3m=x z?G%9Xj=StjTK9U?_7mot{4Sa1p7?`>4c#Rz%vK>fY)?#$6Fm0YXl-t1WjyWVXA>G!C;hoTdpiS?0qfOUs0ZF^R zN?v0mT{BorD|@70Fft;*{eer6fKhB-UM#{JR^vGvK8#huJx=Loalp84i4Y{}uIyh; z>j*}k6N=Glb*O^j`Ei9oR;SXNvztmXL7DkN~Hi2qBj{LqcbkeK6IbRJEEH4 zvK#^o*CC>{N9?CdzTcFpZx8%b#{27~pf@1$c+YlKY)cBnz~LRMs_OHe*!{%18#V1# zm@*qbzar0RX`gyixG1#J=#jJwItr3q>lh&;j(Dz=&mB@lRy)NTvc6@WNx!V(K%UT8 z3^PC;Qf%XqC7RwRz|u?pzO1UuV*89~-4i6b6lH^o-dLJPbCuS88)bwJNN`QkXcH|T zQ%h4%iKT3il!k}D=4A7!CsPn&HH_BlfU(qd!nOzd9+4vP~h$E3f!kihconZeEhaZlJ{qeR2Y`A`KDVv%UbGGyLf2w2vu|$$+CC^bG{_U9e;MZf36ia zGy5eY_2hC{$?FX8^HnZiI0D=MIUX##g(;W_Zp%=kMxaL=s>8%@+A)y#-C2I^sVW)= zz+{JpL}9bt_HOT}V_8f-;FCG#TkAxQB8=`7STk{slqM~3zV&fS%WB0C@nc%oGFcZ+ z0}|Ri39u(AH?fq-if;&ve8}WqC>D4 ziWb86QvJcxq&`wsXfoe{_G~8_2E(Ub6#>zxm-r~-ID&`b>@e8JljJgfh^3-Ba6c0F z=!_#+kra=}dy7$Y+TzfNuD9~6lYge;DT31wj}HmVbIa9@k15`%^XF%bj?0MDmtU;H zRL>0H6)F4apOro_mgixW7s!BN4p^SQiA^Q=VGW6N^}*NS#kYOtPenqq;pQf7IbJ7{ z+k*upGf>Q@2jod0Lv={|1*>Fm=qT3ZbqQ;A%~^&mtB_u2*Wt54347MSkeszF^I^+9 zJ-S^rmkqgg8d|RZqnXM9zX%Odui>{b5HhOQP!O5gP{1qWs(@PEi9!qfpW1N4{nKT?z=g=KoTeofY0a9RUlRRQ4QoKDLHjrZ6w9-P@kZ)ZW>~Sm& z6#P-HSrso>-2C^#G0>^&T{@1jB#}VV2wk~his`gDJ^W4L;5|{NIpgHDGCHqi)}zV{#HFqA>TY7__q$_(7~e?qb0waiGH(&r zf;o_!+7Z2H_ep$^SZ_|G1mIE-LS;xS6^(6;KL+{{jDJ$O zxe+X8@E)6R2b|Y#DSL?!)F?iVoDSUk&UBTLIYEjl&lmO`yS2B;w!If<#nZ6#X&Ox{ zL`;5Fb2Y{?*g=dE|9@_8H!rb~5V&HQ{wa-lpJUG#?11|W2!x#_iXWuUO;i!pBS>Qa zl{VRd#XGIn@wELP8rxeyQR97IIscoYr*{A?p!8g`Si3cHUsFcoH^^riMAq}0bf=-l zXnuK8jFI#o@4x4K>-?P8N7xzs_719An-HKsL=)6NC3n8>npRx<7(LFg#m3m5^A$ ztf?Lrf;qa_Y=4>JBkktaEakr-gsht)Be7p+OW>yQV(9haycdhVm3qje7sID@d^WAR zavkO(eNDwd%Ofx#Qhf7G^d$EMeg=KmNN_3FFc<44O=%m8DJ`&t zG6AM-tTUaOfkjgH^KFjx8*~$DMI;OTaqrjY*=S?l`LZ$umHJ>$x#4h(vMO(FWiwyl zKu8F|8<`@Lk|~y{-QrWWy5@$dj45OM*e8`G8m#X)fXz~U6%r386ZkOv$6Fg_!3%NX z;lyw^&+kR`^#6HIr1=Y>vGN1gI_l>em0!x}zhf-J=Mm$Q)*%2%v~A_sTiXwOxR|kL z9()z@g%+VBj!L}H$-eJ+p=Y=uEeS5!e@8UFR&og!(f{4xUNpg;c+WSQ>cI5k-n0<+ zKmytvy>}P&bo&yMPFxtO&1*?;>K${tcBaAvD1Lj6J?Fpnr##gCIQ5-9C50!KhsDHYgp8effn%o9oP5H)dZWUCX%ahYR4zV>j~N0OlCu8cryC$320!gH^*uYyCl-;zSv5l^O$-Q7|TGCz}|xSPizXP zT5!An_7;Lp@cyi_U4!BV${K6wY3G!va6ROCnkS~;$0PDy%xihbp%zHJv@b}fkt3P- z5}vlKuHan}19~vdQ;1>KP_kF_{2a=Bpwu|y zm9Y6SlFBaNO1HWr}7%!-H#Z=Z$6eEh5hH}54bmWV4KeOZt*>-vXVG=_OE!`xLuGOgP?A)*aY7Q$&y$xqqC6#3xm#xmFTC*c3`0-&8@^7 zEo8R`{ZDbaGV)owV9Uvt(Z!Qu2yEVUp#MAgB6e|F5Xxz2j;0U$z0+d(<~^H!D3 zRe}<$B__V)BNpy!BPOcON^7Yui2%-9S$#Pr3j+GaFH~bl z61HMtJOw9QA>j*WOm-q_c8zB&;QDJMgU)!VsCZmCsr%onbm23`3<8$dB);E<5gBlM z07F2$zftDLn#&)+PnA_^1&=MNGbOXdyyN5v$e_3Tl00A?Cn{n&CkG3=&!3}QC+HOE zLGcWE<)-+mfw*wp?H%s5PX~V9V-CNsRh#pv z69hViQS3Y0Gd!ki{GBz*)PmsdqcE#Ihrv?{+|_pnD;4U>G3V&>8x#&NOwF4h9@@H{ zk120Q=vMAlY#688i>nQj2!IK3LGNxEb`#raFz}#AI>j=qwMRvwM8PVSH!+DSb;97` z?hrPZ%Jxlll%cw68oF}ZFmX6PM*dT}J=Lf@((el8@zCfbqC@u{kJQPBW|ei%V2jRs zMrzn3APC^dOS@we)RnM{lTVO6ZcOz&k^kh1o3|*XV*T3A+)Qt|g@Y|L4#NVK z3|~+q0mD8};iLH?h#g1VKSZD?x)GDJ$wWMe5`OV$fT>DEmFk!zlWb4~AC&rup6m_b zBqQW3XVB8zR9pEy=}&@~r$rPq$SXF{CFsM0?TQEG&f;}lkvF8 zDu3JrfaH}0t)X&mTmFlGXin**MhsZ^m2p4zh+^cVmO>x? zvbT^$D|;#9Oz|u~m`nAM5R$VY@|J)N-GhnjNXW=_CkRPpbmY)!IWSyF0Eo86H?p2g zOtzcjsOU?`6(yTbEK<-nn(6C!J=FL1V*F>(Dk_<*YnQIc(GXU@w*<{yt4X1P%kMdH z&IBv~VA%5SUE>Gn8@O8i?u|(g=hlQnUmH~f!f8b(qtChym89l;`Ix>VG&*Le97_S1 zt6+dO%;MOfdkqnq1n^6bog%^8L#3%kq-6tL?0Fk3milA3rfR*S3Zb&fMwk~@tN^P* z{GJHzWp$taa{R|T=?ZExRJD>)%)udT|9i{1FPrELF9(IE|JfC&^%xv|AZ+~iBVo%- zmJUl!#y;^k1ZKawlA`QJ^Ri425jp+LmO3a)#HAC-xP#yD&J&gO%ks~2BX0p;PK{)D z!lPhN_If+v=0e=xL+fsf=_9>F%zHnH$t-=WFO@8e1nyt^I$|D)IT#v4{JOiqsTT>b zCmz)IA_7+cR4${>&91OSP8u(a^imH%njzP;&%93Q@9kKqqXFp`LiR@~PlUzmO^r8k z(%5S8C{gG9pv3T($$?4oz{)N)ms4tO-;iLNxJ`MA4>7TK9LBJGQ!rSNiLsewEAu~_ zAN1WuVL1&m3&iQ%0Gx&6UChgDW%_r^i<`qR=dstSw?o}|`yKG_Lv-3kE72vu2ytNItSnLT5FXN*}C^l)elob4`?*RtLlsA9L3zVw>A z{ban-ti?rDRy3=xH0xw*+LgzFx$2hEZKbfaBy;MU5L^0l@ss=Y zK|gPffss}7chsA+nKi`FDsVUp&<3g0J+=)%$P4gPHOk!9mkf2e)Tr^EZTQ7C93+4i2~Vuv^B?D)l#94b@xPARvgub)=>|VT*yiz;g0?TsiR@ z?@wOVe{@`m`65d-23<^1K(FN+8E!Lu?0-=7YaNR-{(BE!LtRPoF&3K({%2ApJ=1D_ zNGP)Pxzg@pJMc!ADJ7mo>ch{Sk3 ze*4{l*5TJ!2;SC^4b{Q#-FgD}Q)JM@CgCXmZfQ4e^&+07U*o(FpcBHTM(Fx2l&LcF zu17FG>{T_2wl!r(^a4NWh=B|$KxiW2rfdQJ?!^-Ey+bp&U}f%8(HIUR9W6fp);REE zi0&Y=)*eH|w302*E-lD7G9LUQ8c_=xG*(Se(8Bzopov?XktvHaCbEPBA%0A{m^MZ;(`64sSnKFxl|bRlxgOW9+^PXYzCFB@PZ z6EV~3^AQPhyYjFYuBg8ZKQ~#us#`EPmQz~$e(|-eSHmENBeYEWpZfm)R0?Oi+3Zjo z1&O-EE@7iaT72H84{~vf`g!pFgyMcvkEJE{{Pv-fGm$F@+%gh+5|9ui#(fICkt*{j zjBf4n4m)b0xNg^*Sx#I>v9pr|d1Qm0p+M6jt1Zr@szE#~znQUzN0SDL+Mn}vfAX(K zwq#oNH;Aawj4P54&9-3polT}mABpAVMdUpDGmu+F(37;4tk%?5e6DE!&oyk5`R$TD zM5f?yY?K_04=X^V?x|8H+eb&iEloSsDXn-~>=6U1)eYN-_UZmr?uA zWwsZv&=M~)SLVJ;Rzr}~GdK~^n(B!pw8A?USGM^Hup|3|z6>O=8U!<)WU(06{|5|v z54(s4ovq=Y4}G!35ihbsug(J>NdWecIk)3-JvM@la34L(CEbvMp+jTpXd%h8pCmWE z-mS2bOv+UQ=t!BE;Fyb5i*+^%j`^3EZ>or?t(yn96PoCMYCNxgp*P_}GSZ0w z#@&TErnJZ;4(%x}WhJDNa||l@e`b-4$5vHUF>o7NUAW^mR~pfUh-k-X!^uj#I;S9> z>%^1!=S%+U%l$Yir2)Lp!Lu`jV#QB} zFlvE=sTsz+r=cw?QV5R;d{-imM-`7 zff}k(71E^dcb&>JU&i<=U_v$^L3VKnL814^r@n(GG!FV`L*j;4Njis6x=5mylR=a8 z+pqb{4B@_>4$pAvvscaozk}Kt-qKV;VADEstQASHgXF6fbXQ(B)yyWJKz-MV$mfA+ z-&K<1z7>%1&Mj&!dzdfHZlf^#^ec0e?7l}60Go@JjS~TVy196#?-zI^2E}D(!YMg* zz<5l0OkIX5PxIbsju+KFSOw7N$|qpLS~=)K@TQ`zhNAb0)D`ls7J6hpYkbZe8f`K_ zAe1K(p3Uu5^zRqKjFUezQ@+2_a%wg$`q+g5BAI7${|*)fqb(6m9ZuhPagW1moE!f6 zpHs#4<|q6!*H?>uAw;UhEK%?}`s0~SUuP)n1m4$|x21>Y?Ean0;m$0e%!A~YVeH4( zkrodI=H>F*Vrr2$l~o~c^XN()Q8k=#R%i3hfcoB6b!;)V+u9A($#VBfQwi@iH%k5BIpRa&OWl8(;eqN6_4 z;8aIx1_Z%`?YdAmSlkT74uLU-J=|7|+*U&yd$Yo$K;fwlo`QUo7WJ+zGQ<>+cMJ6@ z4sv|d3|1@8baafZBk{EwZUyS!LR7HS;W|i>EGcB(KXz6-z8YN)^Z2En^1O)? zm^?I(whxZ;Pqr!y`6eo6`Ex_2Sjkw0d{hJTM2V)@c)YdzhFWVmK3nZF1P3G?#pt@^ zVwY5Bn&?-N*fkrW1#X@_36`G)Ce_t}iSN^EcH-04;NdZx0T~^ODGiEOrJQyO@r@4D00U!}0!bshm>@-6`h~dk$z5U9W zN{6kc-~Mt!v?5u7Easo7_Jux|(bZg6Vm+8cNQFT*EV2gvOLlZcv@y^4?J2s@-%nLJ zDFaA;<}0Rts2b}wG*P4GPbhL(p3&4s@RH8ln%$!9%x$ahtK;nvHoqp|+kZe)w()d= zFe^kW6(1qPQ2CQ)8^&~qN;b`fq7n=LLchw0! z_sOSTOjDXJkk$|;OU?f!iP0^>?h~}=DE)%F-p1Qn!RzD;v3c+9j^5A9c7}b1O?s&Q zJHle1Ayiw+{7j=ym+13Mhm22u5&_2hGU*H>m5`$hBu+>PWz83sGQTN)ISoSO7Ph~_ zd%YD8%rVUgtn?IR=77R4tE3s7@_%4XbFI?NxhVkdJ@AS#-^8*woP7S%a#8`mQrFG? z*9#zfe?Flg;umB7^QHfH{=0eXi-lh)fI{DI$kzI!fD?te3)yW(vb`#t`5YbH>|@m<$`F@rdGBx$C;Z(a@{($^ zl@E=LI|-RE?7n{Sj)zR;QRT?Mb;uM?DNg$Uqlm-yp)!9EgI34A?HEL=cF{I zdI#UM)C+s3*E8!_obT*fw1mYxCGb*5b&};U@Oopt@2opq)OXdFJ)whgzp-%PA z6BnVH+hyB&TP1sHwqb35b5axl+IrsM#@l7t96k^tR8#nCo7~66I%|zmN*egqU7x>O zS!;zuSuCO@&-3ZV527o%;|dy|vE+PBV*YjMt#d^;4%;UWSGtHS`)soDs=YoxvR(|S z>f3XcOTTk2n)(}Z>Y=OQH+g(>z7JsTNzqU`eDQ^l2Zm0zJMuCkN5ZU!>;KvlY947m z4(Qnea+W#3$5m)jezk0NPd7*$(`?}1V7dU%N$A0>(Hz86{W@YZ=XSP?yrE&+(bAA) zc%}a}^kR#4ut+o;=u=_z)reThk%|aRH67=l&4{@BL~qHn#pfUFCjTHzLSD2Du&5+k zw;~+vs~!<8B^L{$X;EF!AMPK^$Hsu&ft#^gK_ZP+i=nqX3KO~nFBbsOR5uNQY6>+; zGH!TD`PUuCP~fey)3UfZ4>_zke+MMzdafJ$NP$d~5A=ow7tS-CCTu+A;?6+s)uTd- z=_OOZPFhyh%tU+FvW5l~!EZq{w!`^_dE6E^ZfEH*RGmq#n*Qqpv(#g>c!Kh{1a`oI z@L&n=s3&En4Y+1-V%Ub9YtS@a1sKm|o!fzM^;zrVpmEX-E_ESA^=dTM z3W%DWpBCam11nB#bDRX!;<8GNa zm$JwLlfxklA(;&Y8zS9C>xxwuj)NPRE}?vaQiIE~pwu1Q%;f*>oRX89`wz;u}?Ce1N<;*yD-($xa@>Ak%x^g;+{ zGI=`6k*vw7o6Z*#?MdIc5_DtXBt^YFA5t2LV~QiA8m{^WgA`l+eVaDNY^GRi&PQpH zMCGKKH04he2~{sw2^U<-43=xY4l-=*xdP;BG;m@`x6^E#>JZ; z5LAh*uIE#eFtL-9i2@< zWf&QVe(B!68X3DDsQ1t!0k^F^u=x+{7dx5lq7#oOorYpuvc7_FGkk#9s{&~*l=gwv zPCRYDBDy*Z;tcoIvlE0W{5HS~S|B@tFEE0I`HT$s{xsGMcd3sx`3%V&eR6XRV@5=}Oc!1$4Rx5?d}N@lU4$7&Pl{i$k}h zp?3iqLx4b(e^F!Xo6grgTG!#TM=e%(dm(Q7UwBl`>6K8?yu9ag zifUD#rXDs#7J!+Zt%|y+A<<^1(C0@L3V_1HUrivBIus^zGW*#XPL{UH!u}=svC*u<@*O4!`iA)H`5ppC8>n@Z znk&H@5g-V-ipfHahyCsU2?i;S?8>o9G6fYcF~rlLImxR9=PRajl%1orhY|Uh9g^vt zse8RehQ>vXEel%5P8_x)0k@Gq0qp@C=04k_+_JV)3zzakct2}ti1Ia-f!M*HEc}9? zk8Z>=Rid9q=IXBFY8+Y$i-)EVSv=CI^-C8c;T>*4ZmZZ7^!&_L6Wg!L=xWj*=cBjo zVg`)aRMy(mkCEFhL}R)jb?R4 z+?V?VopN<47O$-+*I6;U;wBkUeCOG@htcqYa(a!sP)YXT1Tm<(9O~7#L&s-ifWOLc zpa?q3n`h-XSZJ_GJ^(+KZ7TLW68?37Pt4gl&;|aQhFebiT20h{=SU#PX0~? zLMaY?8%X2+f7ow$P)A)OFsaUgA9^2gDOqhIF6LWAImH{MA zAmJA;GX&lVjz>PH&;(io_TsG@4hoFo^dUK{&yyI7u`o^Pydehv-W$vbkT$P<()NCo z`=vi5%{j(p({NJ?!nqlK${u>R@t@_=nk_r>U6wSJ|?EtdC3=%z@{>; zKIevAr5s`?&Dv_&94DYuj=+1fOiXX+0~Ha=&h9;X)j}*$a=0?$r`g(atgIS>zhB_7 zKOc6SRfIPIaZEIOekSzoR1~}_o7bgl)9;Pi=gaaY!XT=@R27n{7VqvLd<85@#zjeV z?)5Zdy{1HfVLW4qW<+tF{1dnHl1~K1i zFHjsxwY?XQpq5uXQ)V4_Orjb7W`Y?F;uSy0VJ;M|uaH)bm1a;F$1PRg6@MIY5`E3# zgF;NeUbAJ{6|%1oAnWQ}N341ax7M*T2Vv>rD*-l@zULr3jvyNu_X3!2C|UzLiN(NX z3=V%};}C%DV>zjsG6CPx3KrofZ@fOR6GG;^>tMZLIh;7J^o}rzb=N7w+Y0T33-dYja5a zxD$4Uf;0d(112j8g=FjD1s8V!XWL`Y!5>zL5+f5~YekYf#zaR9(J;9Uy&q}E_XB@- zOa$@l1gH;>^-M`o@Xs7sIxeCcblGMA=-waUMq%n)jJ4dw@SxpZ^UTK$RJoNmpi>Hw z4z6;MZJC+Z|7^lW-9r#4jadZF465T%%5EGBrXmB5gyNqt>+!a7!+53~g;Sh^adw?o zWj8@>YtOzgOx!Y)ASnZTL*}OthuG>Ae5HkYvm!vM@u@t8|IP(RRUfp&B?4vp657_U zvw2xGjoqK4tJ(~wrl)8LQd#O2r*7cqkW0jj1iR)`Y2p(J8D2v0Da8k(M(tk-*|8q` zvYZYuKD4>(dyW4~=CjRKx8TEDJYp?TwGq`fnq4(fyW8PJQ_m`HE2_}rPJ1OvGOMYh zp3rQ>HK49pDXL1IA4oxOnKiXof_Lqiw`0D~dFT!@Jh!rWepg=Oq?7bvcHk|uHC|Ex z^u8U7{VY-gyX$FWxLY#`(bhXZ)iB`?<}4;9UCXkshgDcn%%yZokf};3C+->Q{kV$s zAz;5bKEFT!a^;}2pG93g97eJS19X!3pHVDt(oUSCYvw>y7C+66rAo8kj4nzzbu7y_ zo8zJQp;xOr%e*zpYRahuEecy@4869B)Ut`2wXNo23cI-hRQEvKqLnE||Ij%S+7rSi z>rbniyZr$rRL5mU^Hq)F>Mn%TkMs(%ckC`f>!eS}s*<~lVbqSTDr^*&8hon2sLgn- zA%B0VH7-193-vZq?@n6P2j=OstRD@K_leo18Id*0m=AIWzv2Dv+O(VKAKd{jv(`|( z=g#1Z+LITCeOKnjOq#Z_JLYVlFGelO;c5+<-7rDtUTC9uf7Dxd02#6AMDqxJ z(MC)2QGu0ZCbC_sq9@;9GqI~?50`}_$PA9jvinn*_)nlwbd}z@fn+ z(j6hkh#cIlf(vS(uP@)Gr&ba8b(sOYz0uqqd8J9SkZA@OcJU=^owpoPNh5-C*bbR- zN)b33#aVwbe(0X*bxOw5%z*~T=<|l;%F)hX41l}?=Xga`4h^ONAthMHV1^bv?Zn3`RXjv|h0;{2*j5L>wqRT&Obz;VH ztcw~>X4k}PCBK*-Gji~E4KRvo(aC*quP*Kqa7b%>J#dZ~Z2)+gZsvIFW9z~n@?5%H z21atKbj+Tz3OURbS*JH6IQIjJ!wGxfJmE{JT>a!@z=Q0?7{D<3Oi38c1O`-o8aoNaRKI^u3dXorTI4%DtVJDt=v7#Aa72VKJ)2%% za!CeTb=QTaVt2JHSKXl@3mpWhCys6NXUGF&M;1*G=6kKIf=g|38g42ZD4BUKY)@v8 zwpnCZGjyh1wh**r%pWz{xtHY?Ggd<&7-Or9gc;MLBHBXFYwDv8Q+IP0>R}{3nC*Ty zH%okRJU53zu6mJ$6ix_S5^2Ddte$61jGvbfLs0$900xFaJ)eYtaCQ$cJE zVSSo$!J0sWFAp)B4-7!(=WtrMAaL2KR0vOq{JDvuUxfRa2!`_CqJ+`gGOQNvgusD> z?o>`%+R})kC6l1$-FyiD_2nwpKkY!2?7*7fjaBB(=jIj5-D{mLayx4iz;H+aqqn;N zk3LJePFY>4kz;XdBs7~ zACC|MZvm6m=%S%VG5l&ktmg;p_y&;(2SY_~Gv$C?%br-*p-4J2 z2;kDHJs@mMk0(jqsopZtOI9};c@IkFIX5nPZW$KcXT0d+Aeds`;f>a?+2F&T7$&?o ziDg)m&(|I`5lAm%x$UMJ_7+=7;5gN2pvYFVx>YW-miP@20gJF4AOge5V2q?IL5E$u z{aZf>g$XTl9F2U*nk4E3^FRFtA^<23HVR%|6Xn^+pdNNcz9%RwJt^CxEn(9k}5#!8QSOHv#ra4&quc2)m2mX24 z=Bpe&BJ?YaHi^x3N(8&nzFyq zdDezd@7*du#A8iV+&8*B&#n0jl4q+S=lIVS#go(74xUYL)AdjNNSgc&X~xEil084< zNL4@Jk#!@R4$VVvk$Pc#0vtq{jV=FJe!+h8r7RJ)19>A{VdImqsvlC9BLU@xN-0=} zJ-LH;05LN1dgfilx$J*qZ)X4{v_kNCSyzUKv#u4O1ej0&8|3p`o)GkD<0!Rf$0Aro zP$~S@%~M~(X5fJR0jLC$S>K-&_ny}$E6YWz}mo6yqDfdk8Dkfp;Y!hB~#%TGYe%n9m5=rbDkgpK@jE$o6 zrO-eUmx{({(1Iam-x2%d(rv~r4O}dzGL_ygkiIA!{K-CTEVmLZ4R))qP*KhPsYT5v z8TT^D1(%l%6_C)x`UC%Qvoq>siV1K-2w+`?5sxCM3!on`bHAHf(W*hHG81I zmSA6#Rlxp%KlM$kW&w$dtqM!-UpPMn~e zvstj(egS^PE4mYR_1(F|OU#T(+|fa+syOq$6>daw_H-h4;CZ{;Tmlhj4QCB;!PrXa zTH{c6z5J9Rz8!VKm9LnGK-ZlfdNaHzC&|AHV_`9ns0d|<4U4=+j{v0p*XFz~G z;VBrwhXMRzSGYIQS_B~1_x6Pt|60Q^?O}@-EySWtN(j&UrD-07ggNjS9!fzPq z!Gf10XY&UFzFn&^m5<|*IFJheuL5TxF$TmB_JYkt7uU2BFtCSnyvMe%pZIq$9dhvw za~j{N04yW6^Yc~mq}~zMT!6^XX$Y$XXOVfSGwa$8xY1(B%3+jVKZU}@4X%O= zdKFEEKfahi>`>GgD!~(jzbcJ$v`pBe*>KoRA1j@7eR1Mg|G$RrxnM5oyPO%{A<=UJ z5*PT_8|cS@`|kg^E9+_i0FIX=G(7?9C)>QQVw(mO>$f-ciVbmIg1A<%fX-`Xtd&tMa-Ws{ne0Cvje-)1Uu#BC{DueFT+Tx2JigCgfbK z-e!mar6NEJ?dq%G^&+@j6yimpQ>|{xBdqCJutC6g-TStt-Q_8u8%o%S$>`e9j+S3x zXaitp7+c)gk+h^vrkPa4yLf0zld3L36wxXq5@^Vb-tY6xQAh4|I2`3TaiYwG?l)Xe zjMw3HC`JvAb=*{5aO(3vuFi?CbTqfa0=EhIf8)W)EC`ZOPWH?RpYx@sJX#yUL^+T0 zi|V}&p>UOEd_S?_SHE(qvM=LYH6PC(9bJeCar1hDli^yx+gww}-?DzDN(|kPQe~zX zH6!~Yf4hlxZ5Z5xu>L|yYflZIL#D1OPjm0Q@FsV@f%8aM3I_LDpLvu;iB^0CQj}tz zG}z!h8H%c)v*>vIh!6=vGe=WK7hLMSt|O$86-AaVS~OK^yeKvt(*E@)NO(F)^UIt$ zX*yJ-VmSFO*m`}7IpbX&$5Go@Q-+&2giTmE++83wVEMX4&CH^>+CRg*eAa>YG)K-M zOq}~g_8TdaR-T~_lDLnp@YZMn^;*MD+o_tiZyH3J@A={0v2vGvdGhXI`n}LZrAwtR zdO9!F7MXc5xOX6TxIC31Y}C-IUWl6li2O}27}k5}{#Lu=rnYE0={<6SXAWL+bNDfU`?m3Inls9$VcrtAAM)#Qlu7JL-sI- znwOvxcE)Fl2?``&iGh@D0HvnUGPiFBlw{9_Sm3YJb)Wp)Nbv%=a7-#5`ss&?8b1Ug zMTz5}Qa1YB-n?&NPDMC|_hr@+8N3B25`n3c_aNrIV!@#*_pt^0|4VxcY)Y62(fusW6z^_Yi6FNbM4A!1IrA--rjsiTXDC}I|%6uVzDoVZ3C-xF$us2yx1YIF}RycyCH1W z9UO2JU10gLoW(y$RVm(SbN`A-4y8cU)=D>9fKn}&$i0lQPCiH57qX)1tSPyg*eW@B zvd0fm_dO~57yGokr-)!bu z6^RQWcv<2{3{lrHGgL>k)u4L2#1Si!h;BF0CpoLA>f{MDSyky?U1;zR%j=5*-{Bv6 z=Cs<5;|8IIouz9vkZ;fhx^`IvHgn()n&iqPPEN{6N+&aS_-Q5H4H|wkv|&gpaV2Fsntw zXwW3}mZ>k!L($cjp#!HTtAY`#R82#qzhlqwT**0jg;^V|N?N)q zxd9KcSF#XYdO4SOZB2^JoR|6PnvkK|>+!L@PRvTiGd}fs|5%F3xtad;Wl`<9AI4Ax zpUV4rTaWC4w<)a~TshdD08DysAB@-MicpW-;ruk?@A@t#)EDd95)W?)?>!U#d&c*x zGW5AW38*X3+q61+k>d+X*~b3o`L#gML-6fg-k}-^!v7bpSlS3R6nN2piPoqQTgRUfKzQY6C>86 za4XEUR)!LgF8Gx#3EVbUiYIhO!YXIQ70Bz#1RS% z@8!cD=3+?LV_v|96J4miu~b`HJug+FL$Mp;9*fG;LRTjJqudf>P>Dx~bZV2VK0eCK z;GWsvN@;Dfhk+D)TD2f?XBOv&SbQG$Gvve5fV%Rg*`pMtYt~DLG8-|LX*DVu6$cFf zf|O0RqaB(_EpArk&PBXLunsmup-r%g{4YOi^7PsbW4QeyHo7)vVi-tcK5H5k@SeA? zFYW3~74f>K=>TWzTuwAyuK%|y1Qt4{uP-S1T7y7;`HS**;%@9BB%6fZ*dq=#4p8jg za%V0zn?h$4V2`{sLrB6lyS}J7qH;16^n~dqs;w`atoD^RxYe3{!K(=4^nTbO;yl;n zM`5;H;lL6q*?%Bwh)#C5mFJ}R2Wfw*utNniT_!~2zg5H&WJ;Zz(OOh#3>6jkFKI-S zCE9Nh^oc_?@HTd-U=g0n*-4O&0*@*Nj&g{pJvBjV{(+Oc42g)j@DGfUH^NrmJqGGH zL27rbXT=+Kq5y#H4Mr8Bg>lG{Rk(5^BcLV3@ObG2rdVyBg|Wc#@2)3J#$Llj$2nYb zaJ5i2ym+GMIP>$%8dm(a9-VE{AMEL$67((|K6kl41E)pm_5weZQ9+AiRZzZCWT3%0J1q zieO8(j-?_{ZzkN0wD$-)Rp|tXT>BsAgHT24{i68E(J~;3yLo!@O}}}87+4#K{YYF8 zd@jD}n|VA?Zce&do9EwqdT5%F+sOtF!X*2BOQfMB*WpHiUfVgE zQv7-7m20X!<92+Z%WC84S_ zt2+dM6S(|=LTS6v?A#qD9R1T`q)%-?!g`c~A&R8dgi9pmCIO>8woY6G{Lpa2JQ|Eg zZT*&rpHXBzA!zbInY;}s^ZLc4eqVTjU_rA}G<0t4p?xrN^#Uccl3#p2hWIL4&~NE` z)!e_Jbg2Q(LQL`&gdni^CO@+5o~b_?l-NGax0z%RSaG2Q9ErV*q9Y3k%R3GtSz>e4;=;Ulausn0TDMV67pcSVu|IJ zQKgjQc)4T6+#A(eRr_-F2?;AS{p6soJD}2UC>9?Iv*YZOLNHy9uIMU5^9RW{nn77! z(Z4n-TGo`|iD6~*%M7?{e=EHd6yZcD#>89@d;&f!Aw_Q zDUVAy#FO~^D$|IU7({pP+335 zj#JbaCznx6E$6J+>(EwVKCh(srq|D0Tq)#oLbEtc+rLz-WHICAHPIXZ;7_i3H4_zw ztU3aRPKQ|WPsdTGu>UUlw`2w}UGZE&a>EaFyC|!r|EXU*1CmK-*cqR^a7$iF2;)cJczIhr@)3sbTa$8xPY%LyP21ESd{rW#<-nievP#^32KT2yQNyB0kMrQiFjaBme>yXQj=F+*nOtoZ+H8b}{<5tS8e^|W$ z7C!WtyF-5U}TlMzc|7>oiEhCSzI^;3MqdQU@v7=BJDiSh%Cx~s)kuRfkmQ7+c$ zsVsfu)MVK_rzg!6 z-QF&}?a{+=7QLh$ki)O=09Z@7w8J`?mSz~ycFFSHe}I0PE|ex5tdl_=8_?4E#9DPINC@^G= zfKOVBWd@fvqJxYVS%g0E5EEB|cj3=jGc6`YDJ@DlQ4Pl9G(IhRv+v@W zB2)UljUR!qUa^NYHRF<%Tl_fDwyk}w9vY@wHO&tY;p&vDVve9pBmLo?lG->}643@3 zde`<${wT5+mNoTvlS=A}#5pi-B$ib$yU#K?WwDZte^b*CDIW_P6y2av@UngXk(!iM znT_#Vt-Jj9gMjDCBMs0&&=T;6jdq=voC0jAx7MI}dKQyVzjefz0?I}5vaSI|;y|Z2 zv-Ddc|71#r%7pb(Bm*eOb)<75TnuN<^!;HhV2?}I64_q5q&jhEwl1SmqI%&L9R;C@ zYyc62)3o9V(;$_lLnf+SVZ&_;lPkk3t|=N}x#F?eeWSm`sj>Y(KqZI|G&i8Q03)-c z8LjBF!Srr+4mDK(< zdP1}cRi^S@f#^yy)!$#c|EvbRXI88m$6zsU@1OI)*nkK;{Ha>w%z?G6KffTJs9<6q zmenp$hyw`%{u1<0)p2=c1jwO^v;#c$9#xX$YJ+Nr_6rvN%Y=Y>626kS1*cN(r$m)C z3o{wR^HAgWZAo?viEwmQYgN&!B7ha`WKSYeyd&k>k#xG_PIo)kRix(=)DKM)#h-&S zenZ2EYLJPkv+|f4ou>UoL-|YH+vaV`S-p{xZc)3H2^R&`JKKQyZR+Y0llgSVG!i0@ ztGL8^GOcAe)C@t zD2<3Oyhs6k>1|0B+bXeW(IVDgl1D^$stkYrBz^h#_*Bl}FEPAa_7Sw=I7IeITR<{2 zm~}L$q)sdSSR#0RN^$IpQeiwvw_KQ#nYspR_tGsR%M6oY(YK1cnHjU#%?V0iTyEQf zA}yCK&p;zgC$m2Rx0qH6<491ak^ zt5mz}MM-S@HFf}B8k_9*Oc>v=_+t8M)7ex*LuOO2NNOtO-Z1eg6GalJRl) z*E>6^{XlYjF*tqTJagiLJ^iqP7{>J&V3v-2s(SXa%B+TuA|00OKfC`cFN}7 zUmpME*MA@$11hVWScPbukeLiUOY%3&h{>z@TheY0cIpR(y)&JB!Nqlb$SABjO2v>9 z1ktIi1$!gbV8t)*6a_igWdS+feR3B4Ck z79VEvI3@RP9DUR-S`W^fC+oH?EtG}PY6$xUtAPuCsGz}>YiDZz7>^3^vjiKl3r?Dd ztwO!`QQl{s9T2O24w$`~1R`Eb&O6`%3t-H(8OmOGNo-+D4INZ|@mG&~Qb7VaEqH?2 zJc%e6Y#-^zlF6s!)aSR7JR=*^_*=97v6D*n2|n2f;cR?#>g0D3@3uWjKl=SH!!aoeqRTLp zWsge--S_ptDp3?+d;3f?`sR$J{Slmiekba!4SmxOF3JB)e7?K`O~ZnVO1PUyDK(s7 z3(FWa*w#8>^z}&9!fAfe)m%g);t=9y3Vflal75%cTfz%0Z9(`12Dp4utAtmNbp?nh z8OeFs^swL5@VU!}5FFfMW)fgma#&rNxm5&Zc{}67wZI1xfy#m_3e8grr-`s+5E_4X zF`8>xR*~?+SeeHJEc)MeB~c7o9b%(sm@b-!90*xYTX(|_MqsY!J-5n7ZO4HWJ<`_B zX;S$RSqK40!dpSGmlu^(f*ZQ zjEXz-boRwMq*M>H@YdD4B#T^)*rcKBD6nG_ZQt@SAtD`jSoQ%?(Ll2=sr0|Smc~mf zNvNHddn)-=^e+rIXk^K4z0W4Z`9i7+J0*)CR2?vgcRqL!(SOk&ot>%p&98`L&Cf*@ zFP%zI;%LxDqxB6Y;k0$%+z&7^V%S0p-{!9@77#Lc8N>$^@7sv@T03U@Tj%3XPPpn6 z>0|>A&0?{*k6ck(C!I8Wat~5_`izP~^}x>bOxmB#vio_3Lu^9~lKeXL z4aBQ2fG&4V8s1Ml%DOhyUjR?l3+n}F+RC@yoU-`@Ujah1s7?oLR$n@6k;-FIMLIM5 zhfVvsCMIMwH4Z)?vwO}iyYgXZgNS9B;Bdal#6bm0i0`C0dpWPt7hi%ML0HL)3A0J0 zZ*F*)G8w$|wsb*M@6qI0B_^Q|2MU(t4}L<+IT|9;XkT=1_Oy(r<_UNkV;o zmsWLiZR53{;U2gLmkNE5-*E@FW*2eU8T_c~AXUO(UJCL$LQHXx@10w9p3JqrHO9M# zx-Ax_upB*>7a%doW*7NU6AxqIv6F|(yn4HVTYk_ee!^Gv?c0xf76!S9*XdyJFN3k9MqNN6_} zCtyYd!(TACO+gN>NCT&Bu{lfsXU+j{Rx_?6xeP@)(DWUI7mXYHtpkbLpvfGebAGjA z;4`sHLrz-rwdX(FEjX~i?nr3KiiHvzFTY^$lElNpg-B*c$@NGLSp z&~dtF;WP|r)jD6DtC}g+=^d}I4B7w{BMp7*FX)K;5dS#`e%=9Fp5*f)nV3cFId3nJ zC!eo1(L!d85B6O3Mcb6`HEhsyJ-^EK%+1Ly-!HENikR$tLi7YXORCP=U_DT{ZxLzd z=iRprcNjag%k2O9q5yksp_%bAZrp^(qGl*w{VcHDD*3^p69MU!1CMU|6ot3+X*m(d zHTCI2t935d;K-lbRs3V)9?Qs`E7PL!NlFB0s`>3~BYe#kB$lS9H?JwdIqt{w&j>~a}Y8efW*kJ z%h4xrhxP%6z8f7eb1_Tn>x_+~{!s?%X7&vqwI z7#uyn1sSQi?zr?Q;Xp;oQ2M1_JS9!H!-{`9mz%Ngqdb|Tf9mP?F75;JuG^!eDb%6W zd0=zSxdR?m&)gZ$&|asZB)Km2riCNaux&R(d|c*YhPba(&0v9e2`?IMl5E<6)xy~N zUz(3OG?IjLycareI~L7`)<>k5ZV}i%W1%~;Gx@tYs@tg*3FOmzTeNk^yZDk-SkuUp zQlbEQhL`j;D2!SncF*A~SsoM|H9KxO5}mX1t`~k&?t!3UwR1(EYDwukZvx}N#qzRs zI0!|=+bEfL=#xk3vj!J`(jQTCx2~|3^)CL$F9>I+axYi0V<5}4(S@6)YSw>oP#rkJ zHqj3g#3;A~^;B;wU47F_G<`#g0vY}p>bw}J!b#UNH#dvMHu2=5DyFr^uo7*}JXiYXwdW_#;!(bij8AW?Kh zGltMzAt>OYA4p$+nKzJGMD=q@A7F)jRPMwPV?BDS;e1F;+aED3E!LS zIBe0$9XiN3ABk+M>0-ja`-K>Z|Bg;>bh`9^zFxcKi_l6;WFV_irD63?;b__mJN)Rs zpD3ps!g{>rcI9(UPTgAp%)|erLXPW-I(GQoP~eO&TnLB4c3sqN*gI|Bz-PlCq@iWY zW_dK~FO0@GB4qG?Z8kN<7H^bIeVPuBxBX=FcEk2a5r0LXvnOnSHB6`&rO1c>BCYe& z+09IV7w@2ql!ayl+q<`r!Orv;8S2-JFb-kohbKlBh<@8(?Cqd zGu}f+8PWiayKQsOb2ykDFU)JnbN0Y$L=n!y{?3e3*(f9H9VeP~^Wf(X0jqB30yQ%# zB1D;~1-p7D@}7PtgG?QIE(5{o&o%N1dO-ZF}+roG6IYx}f{1isB$4`F- zz5DZg$)q90&XO3OjJ|zcp2l!z!1S0sM6<60{N{WpZ*baHfJtVA9$02|qvaToTs!`> zGygF_TwZ&3gZ+QZQV42F{s|Lw;svqcD9sVW)eWz98{Um>YTh@?SU$H1VMpVS?cjxe zU!-YVbFWtXS99Wo2ay5=-p_y5d zR?t|Ll~JuOFXtBnP2!R`LUm;OnBD1Ch5@$i@1p1Sp^D15PZ~9+_Vgpy1NMVs9U(e6ow^9NN zkxF-6+EB||vEj6=dqGtUl}pftyNxjnHez`CaXoQjtbA&|^CzdmP`JGNPh7{v*R3<6 zhQr!4+ep*P_|iB}n4#sF(OO8x6tVcVeqb!NfrCP^CajetFiElIhu3@WLqn2obY%q2 zKd_58g6FSDHsMg?cGgQFlF^^+Hbm@7)^h47`$kG6q%sA0<#nvt7DHYUYj;m;yjNN_ zv^oApYZFj`KHJvVY^1>|?1Dd`rtdEwYzK*8@24LOY8WYid6}P7zL-83m+j%MK4BPu zR;rzLldcUtOM*CgR*Pzs7>3@cwX3V1<$7bV5uApL!f1Y z&wT^Tu+qogoKE2mw3{oeCX@fX#enK5o-|I z{?z!CV@+C1Q;)gU2f1gsjQe?zX+N%08~#mr)`a)t4s|U|+LNV}_7QHCV}8MY5($^U zVfebZ?!47NB;!i-OjTzzp=YWrGnRkj4l}5?4V5ZlhAVlWkV8LuWCaOA-h^^36|-7~ z@}yS|^vze3OLKJEja=0@XjQ<@0Dk&I;2u`DG?#bgc87fOJs=j}L~$G+R)$_dU3t-x0`S`pTf@-wPSJTArcVZLPv zanvQJ);wjWAUh)FqbFRy%!Pq$LB0~n@SFpg%Xv>qkYK-G^C z?Pp-P5yg*j>aJe=jGN3Nv)Z1~*t;HK9RA>kOX&8+aPb*v>KJ)!OUimLUZ3CwD4j^x z8;yR2!=nJK1z*RuSCq3tC8DDTQN2w`wx%?WcXIC31^jGm(OXvbiWV1WJpL|1Sd)1) zun|#%$(tWO;ANq_A*AmNA(FnyvxrTaLySsdF++an-sYMIhcN=F#n%j1tzdteUK=V| z=`Y()nkE_}5Tz!;o4&Ya>-L8g<uj>t1YQa$Puvy#!rmf=cN9y*z`%Ii`2yd(1L=pVDug?U#g|k`#j)z6)wDc zxnYiS5{-q$xIPQ~i~gecu!SenD2d(_5HD2%HjLS_b&WwZm7b-OhP1$e0NVt*IlTA+ zb%zGyz0~|Mp2A-lh4$QJH6sCx>O>|f;YuxyKt0hyJDp8IN|;~;vP<r^F zQNT;uucO};prjRwiyD2jK=%r}Z0xPUh4Y6K%ycYCVfW+`mxszDS%wT-rI03$CHBz7 zn-l*(!|!~eUHA~(Yh z-lm7~rFhjtY`nEQPZgsP$CS>^F%60X7}bI6F}aR_DTA4W6hsmz#;Ep!Mcbm|t}T32 zaM(PQwul3b=uRTVK7@#5P4nCiuB74Xtjs_#lCDT8HTj4ia)ovpM`od|bzi}@DQwzl z#@V?gG=9pXH1>jXr1&i0Q0Jn9NSK8s#|c%%?4J4A6%ap5##8nIt2aB686|m~wxA`E z%8?_SEt=vyofLDLD}S?IrmIH2zP|OJR;5a(@5YNTKsaEbW1uep&`{GP5sm$wmi=2nqB#V zhJUHIU{7oXI$*7}C9S2u;tBZh^cAEl(bEJ~TNGy4ou2Pz_wBp4wP8Yy7)^Oj6CmPpAup4bMifb=+ zALf2se-L_yU88EH>PM|=`L**)B7j4^lmCUqrInvj69idD@?1s?A%&bIx2$B`XILQ}eb4^R zHKDWlQ3)j*P zmfu-yCsOKl1J22*_;SRDZrSfw`hL-z3Pb!l7Gy{56S=C=ZggV$M)X13nO{AcL3Ce* z5|h|Z#{D#&kV1%utoyB1l%asr$Xa0BJ%vIh;w;6;e5EVR1SrJ>-xW6Jv*y568rKIR z>NA`vPE-EK>S9bnKpMst2P^6*fXMuXR`G7A^6}Qo0pgnIiqn1`)BOI*ob{7|-~VK7 zkIh)|_uRRWAUoyY$T+bRMw~C)nM?kJwGN=Rt(o;Eqf`UeZ4&Nh`O&n?q$!-ABf!V} z`m=-KhK^-k3g|zgtSy8ej1(LO%_|3307JkNLT#eOud1$LP9y6Qg!97Ne}Qn^2{RU3 z%c>L6WS(9vSvq1Q#`U{)WQJu6E9a@8B;{JDCj!CHQS2*)x8aKv`pZ65R2r%APK>NGK{ zS|3#j!$CFa&*j+ErJiS04R5e)k0cZMxYX6hKw2NkXX8W{A5(#k88EsZ$uUimST;AK zj}@yCDEh74&7t|&xAXh4P#tx<=h`kRfRv>SG&esX7?o!b&&?xiRGW0{0}hgA8w3CZ zYs5)P!JTf3q%9!Q#Tu?1ve~fK;IueGpow*}o!wf{z5(P!Mo;i9_Ba#j&|yCB8Lh}Z zHc#)eYe{ndJtptep83tMB!p8RD2xcRw<&SAHaE`tRVWcmiJm@5!vRe`ucZ&!|L9o; z7VP|U(>-+A0ud7Dy~U-eUVjSV-`yG{8D}@6w<6edPb>6Gv6#4NPKy1zA(F!#?`C_6H!s%0eq?UL};U>FWJit+jJ-Xi-NYPoa&0{-Z~Ukk0fYfE3g)D zoLZb?tER4n^9oeUQdO?^^Q6i7qQu3#@eA+@AHcxX3w6U_b5 z*9vt77v|N$r>w|;Bx%Gaa|D%cV5-NfipPQ9YC|8hw!I#}Fu2R!;n`Q3BfjCUg^R(~|>;=S9YTx(!JUQO`X)Ef7 zTNxm-_DrY=pfUnXE8Vlm!upxp7SuE6AJLYF^AAP^U@o4?pzG2IfV!6K z7)x44Q-7GNnWLR;7mNfu)A6!p^76-P?zl!Ju=2)O`j(Xhi^^f_K@?Dg0Dg&@J%hr? zBN+L~Fvv%UgayX3Oznx=Tg?dzeF@{$>V2)8r1v~yjoLR!sSwD!zr^OL3)m~#&Z9V2 z`hN-liBlegZiHlz`orLs_BA1}9)!0xe8XweB+;+t&gjBt&$YRFD~7?bPOb{uOAJ7y zM7#AY5}PDql2X@9SK24^^t-R3?WYxkwxU~|R&Bo+H8kv5Dr9ija?BQ@6hp0=_HOU0 zO<$}BAO2na6b%c4B&|v3Ezq_j9>DzQ5ecKq0>yF&#haz#_g@dHG{{&gfkl+z2~-Ki z1+(H5Uvr6qYj_`utsr;2U#^Bdvn}f_7t-*-QoCm`hO@~Cp{}FB&9r*LKb6rHhTk6= z(N>s}>iXgEnJ#$3JGt(N2*X>4MG3&fKc+(S{Dm=AR4#;UZ+jhSTw^VB%mRAHgiIfe z02zO#ugK&36_{SxKE-UJ_Ir{Iz^d|!>izmWrQR76dQwi7v$mhMa|lhJp~7a-XJcTJ z?5<9eK*N3j)oHUXz&A9NQqtxO(r{_xIZx8?jMYp}V6bM>-jPl!jLPV42U3Qp!ydzX zidb6_twzqy)hng2#J-U*t7tXNbn8A0dbwbZBbd=7P-a*1)`_b%BE$$pM>S=jM z85&n{57N~RVt6@Z?De=XgN71MIxZSD5|DHAUL-EuPoT&&CC!fx&PKFUj2AYKymh*; z4`jm;BOWr*>esdYXc$m^1k|PoX~0isP-))h=hQ_EjD9l5D;ikAXHz;^U6t>t0hif9 zYbL4SOqS)D7*HYrOq1G!VIr$7DvL+S@3P6KX_wg;8~gVADpH>&sd>)XE}gHuP6J%b z+vv@MEtCL?fvaUkU_=+G7=RKA!mVKxWP@9wA_%TX4I0C{<+I5FmG@%QTv94K&~p{^ z;xqw5=lZu5wO$z}lr!l$wLo^vlX51uVx_G)GCqKK#;~FI-m=q-t^8zCg8ND|MwQg|ZEpzQJ%9w6H3X1C zwh5h@m-*A0(4m&M(X|kX(_c;JI}wF^hR6TccZsC^P8%K|SM5a}t_CZ`j^ZDbram6$ z)q}IL3dHp4S^6JYMP++!kN1j(?qu(d4}qbsZP5JvSHcP<-@xyL9I&-5c0p+J2!WHv z>?-C%P)r64v9L0we0b^+6slau(QXiiFhG4cIw?M-dH&!&rwOLR?k8W77_UhitQa0J zKdmH(7eJbrz#BioPXfj_UrdAeY$bu120TcMT(qQiR16%Q!|VY5oJu-MZv3=1QxYgN z<&rE=H$q_q(vPW5K$IwPVtgHRRZ|D*!!(<=DExneRm3}wVqP)w#H%s|-M0K9P+M6t zq{^;N1kmvAFCnh!Niu&2+8k%nAk9y94xp*&YcvR!7ud0epdnDxy!Gm6$^ut9CLw{| zvpIC*TZSq=n$mscLf{QJBS7R!NCZ)cg&TxSvi>< zH}91%+e-sI*Z)Uts)}aaT_JyyRQzB1^2hdJR}|Is$CNh@OR0ZlQrrO}D-^IL3KOZs zW&b%6HfsILfz-rH=7B>Q4bu5(M~2-Z5m<9|B;?tt(TKorDmA^?=J#a$4Li#VM)hBE zFZm!Z<=|aNuz;?!QFy^0oBw>mYF9)7Mq%40;)h$ z=7JJ39WK}aJC**k0p@(%_#eL<)`u~%RyYKH9M}CNl*n=mY;!Rkk zw+1ke!A0MZFPSGq*`#<2#f$QyRvEoviXb}I>dk(!9VXg-AuRC1x#pQA(c{DJXwTQM zBfp_@eL)5PTE#Zot|F;F+x6~AvUoXHE`OP8r})?LODEZI@ltO$CEu?7w+b0_L}474 zj81mISFbu6^arGpb0`2$V7|9)n1F8EYTGFkeMoU7MMSB6k#rPtK`kQ+u)io_*Ff`; zv_^o<;8`MlA>fK+GrH)E1}Ku{(5r!RpA1UVbc)?s?SoKUVB3fRk>EaGX*JKi8iQsC~J)90A`!cFAP5l@|?%Rog&{wqJ3S6IskB4K<){ zF4i|z(F_^=ea#qnXPB!m=%h>07#%<;S@!;8f& z)Dkw`6Q|7I7#ORC3rjIC%WSN$*uGmY6qBr)$iuhT;Xl0|#orYZ;J87g-dvHYw{5GR zk&KD80@TBnP^Jiboh-7x;@V8s00?wee9)smJ2yyRY@XLVT^Eo;98HI$5G%x+X*(MO zPr+efu5ZRehWd#A_xOkVySnE0$sUa(i>F~Y#(!1Kp`fwX@U3HCnjo~+lLja{cRu&R zok&KYOOf$u(q~^89IAt3TkRohKxGBlANl{0HjcC7#HsF*kdYC^{oJy=5^U)wl|QpT z%#IBcn_QOO>>#=vecP2-<{ZZaTX4*FkCH?;h4szbu%OPYV!;I5<*Mya*yMTz@7M%* z_zQaooo>F+d=#lzDL4X}^$^Zd(c@~!3&Q^~LV_wXdgUR0csIs^^y6<7wzSSFLVey> zpb%;~JHvEWE*c(1O-c4sw=rGNnPfD!a|XjgxCHuzDk^8;n@&54`@hIq5859(nnK1Y zgRYZ0G;Y$#|0g1D>CfgjR?25;-Hk{h@#aXp!(;SSj~#_i1|dhks%FEN69#lG+rX+x zFG3S0gn^h6yg@5ltg}NE$jGJs)$PgV_Yo2avs<>)HnbZ1B4fkl@^8Wz)W~dO-MxR{D z`n9P~Y))moz+>9>MKOF_f^31KZkcSLUckZ$yD$QGYN65C#lge;a`P$r^toh>x_$FK zz1$QM_1jYUWOQ$ALkZ27-&vn$j%ik z=s)g}jWdHZjsq}#V?a17Jj_Z_tEO_f-QXBHRARWspT?fHw7xx=Sm~RIn%Ps$bO2eE zkMwV)Npw0zx{}*QyRk#*cQwEy3C*7D6#9eI=)#zLdU$IXP+et%)mly}=0}aW#TZNP zIA}p_1@VWx5>A}$IM*x#Se8pd7XgyH4DeAPuoowDb>r|#NEsvDWD^XhJgnj!Jyf+S zf%qDo?YH9%mvOaTmm1B7^KQoVcCmD!fC9b>U4=VlMlF&@+VZ{ZsKo1e86D(Nh(X?` zOIQr`i3_guyKdDM{cw5j%AAXqj#2f2o-&AZ)@{+ZgUJVh|1A3rCv&XUA=s38VPu36 zuYc{&dx>@4zT2Oo%uBgtnAgSH@+$4y1 z6a~u~V`pqj>&A?WVlV)Bea-4E8X+#uC;D?-uS?SkwZg1h&Tzsy~uZloxBY2LsXu zr|}i@SB8n%vLh1_fO{H^r%`!W$Sad}KnS74iIG8Rm8aHSeeGBIHaBl+LV&TDME%GX z--Im?SM7R9uMbk*@mv-9i>81o=S`BsBi#J=IPCSmXEhjStY*)&Ss9T1?sytnZ&m=S zz2*6;Z*V;TmWLT7T2bE$$kVn}a`#Og@g<0+U-`>Pk80_~AyW%tJ}PfLr!Y zkk8n02PD>$GE-`bql|sim@Nb?xBz$@8e+9YfK;!kyOzBa3%*Gk6NO;O*wwMI?_STJ zeNw*$^e9Y+-XyArW+h$uE;hgmxVw0F8> zQD)08=pmW!qkM;90p0bWO=7DAAEVLN+;JXrZ=zxvwkG=hOHl!>*VbD&I(9rD4n#t#y3`k0cwGOq9zap zljHgtqSw(S2SE}N5mXE)%Gqf#ix=>1^N`sxOt1PIdb(>AMU0@&2a2mBwxMau zs>xr2d2aL!x*aen^OYtQ@@a;kovv34!H$OHTn{1*8DJ=NI!4hY5n#HNVfrw~SiVrL zKxoo1acpWSIQSv%eHHT24vdtmUe;ectUMriH(BJ=#JWDVu55J}@w$s2=s=CUFN7kU zvX5{7sg%zA^SMV#pZC;7FrQ!fPgCubHi# zK4{5HE7LM58fl+EV%_DC6Q#_?hXZE%&6+2nqbr|(jL=Nkx@uOo)M-Xr(U?S_d?s&T zV(q;o%w9r1xqy*KwT=8L@IB|_C%MvnMDwG0Au~z9MMZB6W%*1<8Bg*mPgeDXOZLd& zgDS6b!gD7oiU97;d{FRFuqt$$x;l^*Ht}KDz$~FPNqD-3ASca?Wx!IU1&{(8n(!v+ zA0DcgC72N`IOZ;|bcG%mbkxD(Hwl64Fav&jpR7UhfArpKPwaxK%Eb8p8Sj)`>=XCe zqj53iVUWq*0E>R?QbAt|fF#Hm2BLUOst8G4m8&R7-^IN*pc#ZICOZu@Al=_CZr#m% zN04ECbBMrvS~q?Ctq?|!ii^z%mh$_Bo7p_|028ICsQLohN9XnU)>~X0U#mT*I{{1- z(|SF6Xfdo%9(>R$=~u1uk8FN9;2j)uzmL*8j!yoTP(q)Ko2cMO@i9AycW>pH9T!y8 zY$W+ZuW-s>Y^^(M)U(Ie-F{sV-AY}}hUV6>2^4bcf-7;@epEqq5pphPx)x>~Sl;Lm z8Y}Q*v-V&tNtv)I-O5K#eeM)X|87D7Ykxh%;&6}D=jPys?_}k{x2ar$=FXF1CnsZj zV!Jf2i(;58^x?rtN8Dc0s=L9{ONxY1x-=ux-Aw5)MH@>>=-P9vl6)L85&;{; zaLCrl1f&^;30Mzj21uBHu;pOZ{_O9NY(qzH&}?Nt4on+8hu^EIj=n?o$-uyUooXU* zgU%2b$CT3f>M?$~DTO?Wx&l(ktcA)ctFn4vBuACe7PLbEu)Ow<2RP+c^FMr*SPlgo zu`CXx8uK4T5hLeRE!jvl&4A0k$XH3XUICt2f(>!`_1vDw_VoD&6kQZF9z}(k6n!KE ztap++MGRHJ%0xmW8Fql3*Zv6}`+a{;{&R`wKneQBmu*~YQB^HTWDHxch@oN`ji1cd zkOXfZCR3ZY+PeK!b=S?by^_Ad@ZB|~_Q_v)Rlo_W<^V01Hhxy!6jBHacT-`-K^1y7 zS>NKOmckPbRIs(K&vw(hKO6Ri5jABFf4Zliov2wYlqOf*U8;4w-Yx+2aPlcSnTwUPR1@ z#^G8O$-Av)PwS1(d|6yJmKnG_otv5Z>;;XlmI14xYJnL*{)aG50g7&}@VC1yc3lJF z4O+E7+YH{m5e^rxiH%$5a?l4j^Y;kVjG4VKsO`KBO@4r;q|M~-yz7M<6>?8BGmrrz zxF3K1S`c<*%FvR>8+2ksFr{sz1qyjW=34^HI!w*CqJG1|tBg8@&b z663m!s2j@5zcAS^KLqK%^fVG`$4}wp9K5my$S4^-zhMDK&<}z4$>>x{_^g{uQG+Kv zz^J^Q@F%{%&KslVRa>)CW4TTydC? zDJ}qqwG-{=`;!_8a#4^(`(VTl*aj@XIsYIDbfPwjFfUiM%G-hrjC{I)!8op8 zq}40vbnz!K1hWHuBtgA8Sz#cvmku)Ab%9KmM^MCc>muQbQT0TuxneH9Jhb4#<=@ol z%U`aYqNbCXxLY%rd_B3+ZRD7duQ-u&US)nQ0o)V&Y#Yk38M(`V0Bmrl^I`)8w_3=v z@~$f&Xn9jac!U-3@c}3p0Bg&zlD+Cl8a#b{KXJSy#VjZ$;0w)b{MOV1O&YaIO#7%s zEM|GTDE*Su4zv_-h?p38uxwU11GeEG>#|b3=51(J*BljiMMYP^A6dPrkn*a9^>MF( zpB~67oP{)wj_ceWK!+XSyV2l3C2x4_4w!*WIfFk P@@P>#zGt?80000022T;t literal 0 HcmV?d00001 diff --git a/2024/06/10/startup-data-health-starts-with-healthy-event-tracking/kafka-inspired-startup-data-pipelines_hu153a73840b0658eeafb17cde694e5f88_121924_360x0_resize_q75_h2_box_2.webp b/2024/06/10/startup-data-health-starts-with-healthy-event-tracking/kafka-inspired-startup-data-pipelines_hu153a73840b0658eeafb17cde694e5f88_121924_360x0_resize_q75_h2_box_2.webp new file mode 100644 index 0000000000000000000000000000000000000000..a969c3ed7340acb13ac409d1709dad2bf3fd0971 GIT binary patch literal 15454 zcmV-kJfXuepxi?K8F`9A#~q5sKmIrA~+6T3s}L;V}3%htR6eWwQ;|9#-E zm_Mq1{phuc{?oM}w%_-D0Q}SXL)doc-M>5+`|sMXVNZ6x&HEkdRrtPc{|o-p*a!S~ zp1=6KdVb;mYuHQr2ly}j|L1+;zuNNa`*-|bVh)<#l|QpB-an;N8VTffc#@8NzLmd| ze%}*p0mgsL#sf%U4;A&584KGc`Oq0i`7%50eu%A;2`=nQ`A zDgNr}>UqFF_$rw;Eiu3H1Zyut3z79nci^7dvWUT)W%< zol>t*(bKELh+PX)ZS41wiESp$eCg8V&>5mG87e3#CMlByU4E|EL|`kzHb92G(M}`F zDaiS};sznaELjDo2G+(4w(XX_mUQ0PZQP2~@?jHm+V+=+=~+POzpe3>ASdYj2X%w@ zm?L6h0M{{4!N<70MkOLcS3S%7SG z|0z5t4-#^_lfoL3RQ{HyMdhRMD%Lk+^tDW+RFw*_A&^t;XbFV&JBOid(cSRFH`r_9 zWB^OIkB6U0!Mlve$%ggiLPuOoS;+znNu+l>k-0T=4F6S*7iqdE|C=7N=}KHWRbHlH z>4eVFM&E#S$qq9VT^~a`k!0$_!jYx0p)Mftsm?1}y>ndBEaDN4F1$L1Q#3HM!=wqZH%QRP07bsV_PI-mcJ ztP8Yg*CA|?ZZj&%MZTGlN06zls)8ML!v~Tv{Lec$IU?^DeI>-$&~lrngP-uGImk*% zpNUBTIV|_B$7QFrzziQt@Ru~?rlb2F<%i8DaQFRwOvpACa$!_b^yU98+p_{gz=Z1m zs4gJF8+x9vj~Dg;^md2;!724^{u7kBdu1O)+B!7GyptWBF4)HO+9HE57q9 zSJvU9k!0Zg?iqS@Ax%S@n|enM=2`xNKIX9s3N&;)l9aT_RI-B(OdCMU zNNVcl^W^15VZJX>LSCW=@{e!TULSnQhTY4pDph-XkvY_Mw1>{X+ACWHD&P)&2I- zP`;%8FU?y67BQz+Cs-8iYt*R4^Y?+bgbR{?;^|)meMjtQFb|G5R|^_{4wd>;g%pq` zwUwGGQE76u)%jfcWsE1QBcd_)3>`1pVPG?gg zPr^I44O=BVdAw=mY95fEHqr;jp87_oZ1!XTj*s%2_g9vWyJgb9|2f(uy7P+HXPK*S zpfkI6mG#hh;^rOOS;bD_x6yJ4zl=$O{$Yu%(P*CXzKHnzlj^ifRrhXr(F%~>B$b|# zMFQ-XB#&tpc?7F}FkEkHQj26R zlI{=6xUuNuH@sxy{{U?KvpE?8#&^`sNIyFUO_*&0L$(xFpX5^$(y`GaK} zNq*o}a<8vuQ(A{`0RH{s5S*oM(49m)_NoV|AkPbtvsZ2e5=zaGHl)Qc_N{d`O`)QH z?kenod%UkFfglxTDr8)BI(HWQ&#$&E^Ov@;LPv0 zxc*HBSc=YTn!Fsxx0{Jppe5VDW3N!-JReFF5MVyJs>G%l%J}}z*9+SarS@ODJlq)k zt;sopvSm^89KvjtJ!BP!)<`vSz8_WJmsAd}vkcG14rRtAkd^Pg9%Fx=+5y~u3o|#} zVJ3=m->hA=K)VSL_xgZ2(sj4o+%;jr1Q@{QE?6-C-s*pX*<2{3_f8#uz+-ah6oaT8 znOAY#?LdW2rJe^nI~k2_-y&(4787WdFEU0noj0ZTw+dcyeU`6dLRBCDeW>FY&iDEc zQ*G@v;wknSsQ8C;SU4fxjX0g=!kC)2GUDRUV6-XQczu9;27@r`nzCr~yGgCkZv|ne zZh)1>JuPTG&Fwuf<`i?*ocEtN@IrmB4`D^up{-Rm``D zn$}qYE@+9y?j6);p6n~2S#LUGhm*Cj-M-`qQPT!bV=Zo=Yc=5Xx%&aA3FE{vj`)Z{U7V>^K&oA_erHGqdu%qu~OJA5R_GU z0@F(b`W9c`?VxD(_Fe0=$V5b*r~w~i8OQLb`2I_oFC#2{v)tPgLV|^@yZhi6$84I| zAq8r9O~+4$rLmcdss9+@fAybfCljXD3}iuj$R42!D!{FG^BZBbDg3f749|&+{i8JL zFtT z!Ix*ZVllk6+{^&EsH*r_Jph^5?pL8;T8iCZAE7znisdCnqXRidE-l)o3K|j+got3o zcBpP*X(LRLc}F?nT1@3>KaEpb|dD8Rf&vylQZ->+&lU_YZz z^@y&NgS@;=f3mx=1iywRY$LXS*k1n+Oy+Y$Nn_O;A1_AJ2HjEsmbqU|Q=@bll7fw72UB^)VVDUyHb`q94M#{q;=pSQu=`l1Zd-3FsQF-ElDRf$%ePFGCtXs468@CvGN`n zVa{{mlD-}y1KcK8lVzH`RC0(`1W*s_c199N$8+r{)v({&AUqo@o4ZO0Dh`=^#=_<% z1Jn0de`v)V-UfIZace`6y_&|TO!%)_JL^qcf`|czFigUBW5yN`CTa|0>qkqY=HA8T zS&rVXtL7OG$P#Ru<;62Vtb~aLT1`AWzBn^{hFtX$NCA@0({qJRAp%K`)^0!^K%4u` zL5(yV`EQO)#+IyR@H81bvPJ<;$oTZ~H9Ifua*Uwp`BLYFz+Mr@0Gwe zW)Xw0nY5UXgY(GHBnPn6oq*MC$6W|Fp@$hr(YVD{{l-CBPe&}!qKli)_tFA>9iqnX zfkM=6tzW2BhCw0wc~bEB1OMU*N2Q!$ZO@XxJcvToVn|Yb17L!pTiRSn%j;LXkW#!- zvXd?_0jdUEimf}O$cG3-ZGhf>oQIC|nolYAXuN^>dq`BB9BsJ_1(-M~ zqg3<*!cX0x4rhP^VU3{aXS|MaN!q!0gU|gwUZ2E(Sea#W zEo%b6aTVZ13R2Oj!VPBnCr6soSF~f$3eX)UussD z_l6ZHN;~~BUw{Wn0X@)sE$%Qj0G}in0Y+_5E{<*Q3t#81YlRKRm_Gfy+sGxx#9T#M zPK136HO^uMaWI6R|JBSb=vsq^r;g@BPUN3mG`JJ9?^DbmR_;;4A699Fbywhh=sOxM zsB0#Yz4muo9!pk*hcdHCGkCaGtQ;(*}eE&OZ2!HIWG~0JF zd-anBL>f3MEPXNrFRjNZvx`kl#xa~VzsOsaaFH1= zy^2$2Ff!=GoQL11HUfx;aId%O#A&I!cM&#OPgZf*LKqPHLgU(Oe{X=odE|c|C0{}F z_*0NC86UiGt?`+SbPeGW+@OmD9`2n1Tz%fYbU2?-bG)LXONREs-^T-!CFY^@^h{`_ zGUZ;U_?%<;O!}0W?*o@qdR*wpsd1K7YW_RmY4og2nkK6fWLuUAZ*T^5Bk%% zl1MS1u0uXi)<0{P1-Ze`CXT%ZzI5Q0!}fBawV>;)WnJsjrDT?*{J2^S>0C_f5W7*jQQ z&UrNyR{$OTTq={lU86mbfuYqjX;JpoMeZ_L!ZOdRZkLmo%%^r%RA4CU9g~-%V%BvF z-8c$z0-~%m&5T$UO;D4)A6K8Y<00$Lv zP}E|lzBAUW`(~mi1rkyhyS8^}qv%#8Mk%b<=%koCY9qF>D~gu&o;16ZW_>+#hnX;w zarHZ@d!ZrnmnqGwX19vza-`|ZD|Gn(T(z5@EtAE0#j;#M%H&YW;OoRtZg7H_)2~oH zNVjLHd9u}jG+eq$L-T}w8)OCKu&`|Ze{Fl`3{G!8h37%U)8;v?brLX>D-Gbt*3ICqN1vzNP zTjY09;=|U!Gfg-UOs*lqoRTN(-J;9qi*<>HQRT}E3JD zc`}bxA(#%B%N@Z-4wq$8kv99$TQM;HY0XCwJ%41C29=33C&}yY7kXaB>GZn<9?#uN zm33iT<+G@aGF7sFT1!P+OFQ!vP={aR_8;~o)-9O)*TiW*-ORgPo4Km3y%{P~wDSL$-%$NF@EJnT|J@pd|pej!EC?8fGU`2Bpes zT*<3_5W2m-gg#(P$>cl>#+w9Kcx!Brak)Aeq;Z{q`wUv>Ys(E$Rz{4~{K)nY>>8*< zeGb|U?P8}~0c)Oj9_yo|oGarOJHeWl9<#!j`y{$_M7-Gd z_iSf*d==X+jRpqtAn7aVi(3||{XN3dJfUyz+Bj$oju>G<2S5^|Y)mt&fHFL1KOz_@ zh&O5j^syH)l|qQFED=M7;D`v^LlRQ$yd3Q{E_{_N(Fd&~Tw`Ah`t!na<N}qg<_qI&`PewHoU;w34L6MiQxP7&7`~N>?Y+*q^0(0L{w<;Kp+M#tT(QX_dI=l z7`}}Y79*@G%?wl{o2v}zu>q!{qrwLXnIE`g5&|OKkq@+Gen2Vx(b%p(kkLF=7Np{d zXPqaw9CSSSa!@~>WZYe^KyIMH>=Ez7p1Y55CjA1K=;*JTJ~$v$6R`d*|WHdR-jjcB$ z1OcOJI)P3_x}j#n*h-&&Wa zwpg^?^$QX=XnL$wjc-#3x=0C2*cN`TJQ*bZy)p~62N9x>2TDSgDYK~txzDm+pHa#Q z4$EeYhd1Wr$zPNd1!NBrGb`OIoplc5CaZBtoAW%TNSn&F`tD*lpEf&_FtGhIk_UeM zWIh2CS8{QTjA^0M+q44xkc|y_m`Ko=+$!bk8SBFJpU_@o-ka4aV`H(@8Z1WhmRele zzYMGD;V$h$l1TemZ7+m02~}#KIGOHLIQX=svhcX`?|A^Pk#xj*vf5HSo~w7R#L{uf zhyV4iTgBWYl2WfK*NjvOx!X(JylwDk^VOe~_KFP@+^j1Ln)Z_xBa3yUqLkU5(@gcE zOA;1+!5~0RNXpJTKFI}YlKDY?^$tF?NFXgiS-knM&r)m!l$@eIdYR<=mvCip@ZHT( z_s51S#&gO->CDa5&p5_6gwJj`WDsUyWH3Yeu|mW=piR>kRP)S8|9V1fike_)C~1}S zH+$8Z-)!P&l+7r&XW?6)M{%&idK3Kf(Dj38+B4QX;)Xg*4r3YAu;ez^N9@OXY8fXu zcoZ!Uk6tZ*r^R?Zy7Q-ZJ6sy)rEnl0BN$T+V@QJG=2*w$0$9yq6zBPsZ$F&w3jj8E z#MKa z^6+{ypbotMn2wzTxGJ9IsV5=_!%S=&eqbKY(jz0g#?yPl36j^kat_aym53qkdMnSG(Hxdyk@+C}1tlM?OWV`R# zzfH~EDPsnOAdG0mJ`f`c1^ou7J1UN590_DUJ^Ez}4p)h+MTqRx3K>T!{Tg%l-~0MZ zew89%(2FiM-iD(_*_6$}SEs+n<~H*b-c1e*IA=z_RFhlOKqS+>s)aRSBoDE#!Yxlo zJyM45VrR!C@`|JTc1f1q?BYHZGU3!7Y2={zmSC)_rkLua2asTQ=ydLg~GKAv?;C)43dX!P2jeI1Z1p3n*97 zEsgGIq4?%%#422E;v*6>C6>@TMd<(J*&sE}6gzeZ75U3%TH4_bv7ND-zFMGmG-p+t zb1fU_9`bi5K16&iKd_;mfs1p*$LJU6P+_i;v|^7u9A)y5w=(Eg_x8tC2Fxtud7*%5 zqsP|n8BpzAe`ItwR(!x!hngf9eA0p|tjdkS+V0^N$&f`uOx*}Pyp|Is;$O!W5#lm@ zpsZr@Z|?>m+SXmw5W>_ICL76r7u4xjosb*aB4Ekr9gh9CSEb3BeAe+4UhDR4R~5CW zJB*qbS28u5+EI;7dZ^3tVbjCHiiU#WsI37|#(na4(79Eaqy}_X@-}okNJRXroA3on zZhrr}(g+Qim3fIk$iVs@_qoi`bA)2h#b}EEF@u00q2MM%iLT!$it+pU@fLHwEV4}Y zI&0io$f-B<1#q}q+}=EKXTghr19uwJ|KT?askA*snKa{9e&@FzCd2UxFTb6kB{`Lb#6!6y9L zo791(5E9-ul$LxB8EP3=uK9aP7)703BPtX}zcWS<6ToE~Ej!Mvs?rK*%>{4bsM*OxNsT%H3$pQC}2 z1;KF37tomB8uI{n}Yg3DlUlM|M zT%q@I0YpFsiD_1*q@3$uf}+9O?g z4-7lux#r9O-%H-w4;L5pFj!>=2g=vSYUlx?`QPjbn9?j<>h!D}o&HX|i7`R~5$Vs0 zurs7~ESChA3STqM=D+Gj)UP!Q1pWMn`yfn*GvTLCI}_h_1#A2S7W!n zEb)FMCFmDL3bOw%nT5|rZ)?aBXOpzgwW$8cWe8k@BWJCf!D-U(vd8AE(T7~y`V`8h zr|Y$Su4MNx53e?9QShJf!0?D;pC29AwuMq-KbZ-kM`4KZWw72@l4sYuFu?UI&S-=J z1(s8aS+2q?lf(go1F|fL>RrRrtXxW$3N`FDgn7)Y%yOBk zG|2Z8Ra+)_ubJseyHTpr61kJtohp~ivFbk(+8P(+m=j-)yIqu-=EvLxR>oeK*crx6 zTyF&Ge#h+H1i3hW@ZM`HHd$b;wHk7*an6KYTP&w#0|nh1T&Zu9WhSc#ree%EuZpn| z5dpA@bf?gx&p&fRiQ7R!L~S&!2+*Wo9bnOPTo6ky??b0mt#hG}Vwh{M@(s(u!}^c7 zy7EpFj@KG7>fQ@-?dRa($5^!ctR3%_&1t*3!;Ju{=-231*K_>WH zQ>v4nr^hZ%b?YU38uX|cU(X@LaH*>*$q{wjN*ut3{;>M^){@z1;JOF6-Db??x?0Gs z;1|g}w2R{+B9Dl3p6Hos1keE|@2KTAQ>GsroN7G>_Isqp1-h8d5trEwo9l%WCs-5I z;&yVx?w+`f3cWB*<1ks)Wq)3<_qMe<1>5tFl2=+{%1C6d$qJzfbcIjQP-g}DWs4UxZEu|CQeBA{9y@h9Txzjc8F~~7iiC5|`kn>1Hu4X}h z9Wu1?C=wL9WpcdP+Q)M+pKZ%nq~Un$D74Uc1sUe4!0hkiB%?qGQ^fAFo*@y%R9*Gj z>+K6~7mKz>5_w)Fd*&O|m8jx|pv)!lS?>LVc`K-F&fRFZgj6Th!-9gsHVomaCDq#V zp-9Luf1EtBd1g~D9w4d(chX}xOP7U{z;CtkabG4#LFx2#=vx2n$nNqE>zA5en^fv&;nHdGNdP$&NITBawIYdf6!cV0;O2_x=dS3zcN913y-9Rg3zdX9RFEhz(mdD)*r z>HrFk#q1=hOIvtWj2h?o`dvs-7r3f40EN|pTCB1IPYeh!zRkZM)y+AKmn& z>*TxdjvFd216rKUSnUTh?3;j2!k6dlwr_&s464h1m;928BrWjUME`UYa<%NO;FE12 zYM}Vg&j$i~*}ZA>+CXActDl#i#>Uf#+Ug~OUmPGM;aOwh-D)lq<)(ZWi`-Ub^7qzY4wm7B==miFr)H@ zzJi?gTAqY*r)?^Isna6F;d$=DBR=#YfraY@^00lJ`$F$b?Ts>+md;c4?5yKH zo0Ka-8%c34dq{=dG`&~>bMK)dF5Xks|M_sQ6mfq+TQ{lwdu`C2uHOIoK5Ig_EWi4A z-KE;V$C;@VXJ0>pfrh~Yyxn~p`*46psqxySDxJK7A0)&a;7k+Rg|ru>3H4`wawT0qxH9jCyMAQZi;U zr0%`Lp;9u=&zAJuiOQM%E#POHwXz7@GZT81 zHJoriM4N0l-7Nw>(CHB$Q~$NooqhT)wzX;wIvB`cUyR%xtc-sTS_Z&N*j2u1q5LGS&jX{HaY$o+SBKC}0 z`(xvIOYjTs$TJtn8^`eEriabS`G-o+c}9f@ik9F!HcG}WCPMbP7-`9@cj^LsA>?@` zbi-P)wlV886BzJkMHfK4K`K!{>juYZHZLH6LI?)=yh(f}JU0~HEBEuxu+VQT2Nqg7 zI_*y^$LD0TNiC&@->VC|;a1HuX^4gefT~;b237>8K5jQ3y^2l_k*4D=Q8K6q4ZFYwmL_OGVbJ(MEfdQdXbXo04U&N51_!K{pqs3;pmI zl?*(=d-fJohBk}JjM$089P09QDj~mZp7vAtaXW$V4#A(Z2*_@k28|aRJ&SLO2pQ79 zIKm=NTLQXEK@;LvIPA3Lw8*L8JUF?^!o?bD3?d|A{xk!G>{BnOcgdoLq3Q)a|2dSs0yq*;XQ_~O@8W9%c3lSH%~vURhLcqKS+t9b zXr;D@JryCLd3yqM)u7IRd?9@QWRsbab~pZB!(d0lz{7Q24u+p|CU)zw+-p#o*r}3r zOJ&Jah@YSV+E~#>l@N_m;NKk>*t%@p)6;Xd;iAc2?71gVX{=6pLuH-y=7j3Eb^T>7 zsRV^T1t|b5s%e3v=Govcx)TR%l@q`75nFB5oZnIl@@9p0NDn>nSCQ~CYI(Hq&*KRg zs^R`ov9U{IC8&)02|T5Rxk55bSGenZYo@IM=zI5m3-o~nyc=2XdwH>(Bx=i^jA-MP zoHf$tpv5G}hQ#H6pIpxUN#_6J+imdx*^h7rV6zo zfj98?=pwzs@nN5LNiUx{U$YYbou@mo5hICtF@75yqP!gPaV6JJPk4)Lj#xfl+Fjk0F{6M} zB_ZX7u1zltpcuwL^om?^);CS&_3JT7VhO{QvsqZho_Ux7=R(-7!wv0vc9=?0inMZ0Wb-u_D{OV@D|0?#<;blwn4Su!iv#)% zpgu%361)ghc7KHAv+CC3VpPL03CC{dLKfFU3$MNHGNwMcJuy7^0Jp@jxPaLhBU@ng zx40TG>!?H1_kxq+ckS0yFsjrCLBYq-fcsn|G|s~UT+ZYw%!sQvY9Gz{A<;YX(0)y* z%yv!~r=UkMlPm!mjrK8op{H!esO6@O+(c>rD+NoHjw)e&t*G-uf6?pN=DpR~s^Rpv zte9|W`^2^2W`kE^OcQ_wg8NtpnDGmwu_U4n&coz7*U>((2ySWFsr4#kp3_f@-?+|z z%s94Xhb6`$`;lY!#bLjC;qL`x$KfiyR5{xe19p)KvUneqCDa4k=9+K(YX_Chpc|r@ zWVeCZRx&c-wd`r`8_jor-)JD%4!)fPf>i}JvU-Mlcl9+mbWhHzm?o(<7eyJ^5)w8- zVlw{759CYj-9?pgw18-u_T(GgAExP~^K+R+7%8n2t2-T)0ls|;K3Tyj3OhrsX!(2u z(3gaaJ7ci&f)4+BqCfm%qD)UMGBBstW=JcA^-~TWsOqJ$Ug%UCk)x$@bWjCR@2g zWts9-ZGfl~TnATo>9uQb@b*Q}=*v27h&#Hz`WV4=ndEoe&gMa#aZpO}NyD(`tEjAT z`@$A_TRE?Go-b;d1H3t7eYK_ba?lfJ!T=5u$I*?7e_-aq)Ae`~u(;jk7!qVEuKO$C ztI|Z%Ri0{LlD1qLBYpX@{2NvAmV(&$RtRmQ&MtTr=VVt_lk#40l-c}X<+;5zim`dk zFN!scQC+?LFtu-NjcP#3YF?-VnM#!MY7eFd1>*CE5wCVd$QMzqyNxN4{elNf42AV= z>hAAw-oQXBjGgH2pF7WpEhun>fs&0>tK6Per?va`0b;2Oya6oR`u;!Hl!OMzz&gBI zZbcgTxqAkTsKC~BJ=4WXL3sN;Z>;NA&RS|bEW!OasQ9yyedOzVA0{x~s!5+l|Vh{pffiER055a4^vCgf;<&{=7(MV2@A*-(vHD2x;(Q=<^?) zN&s}MCoiWd^{u-5Sts z%W>yl8ubSNxj=(05p-G-el*#XNd3X(W+d+QHcVvZ(Oag~6^6;@p=N^4W7poqw&9ZD zz{DY2UAoSj?}>pO%`s(O)9A~RjcisTj@iq+`qvZ^Ub+z2p72jZ#H^@<@PRd0v~giH zE%?%#CE2uqwhJHceIc$5I$Ss>_z^~iNftKZuhZUj1`)w<150;sOTI1uiR z)oU(Pauv1Awc=UNiEsT0s0`n;fpP zq0yUqI~&HWscNIxhw=f~A6#x)=4iLB$+rZ2W`;7|hbn;|@?hky|BIqFaLO9jK*yFg zt`?@u)RT0B7t*IHDQaxWd{g`U7z>XwKRN$G9BKDo1URzEa1aIiPt6<( zq8<`T2+z=U8AqvB?(Vf7uB~&0a;XN5-$*~kiDez{($m1v1vFNYk_AT-&+5pRP@K%7 zCOT%%om|DG>aS2S_7Ut&d@DPGlP+gHu@p`6#xttG^Cy>zl)p{7g-W+SGNQQ3D-AbA z*DK+>3?%Rc*~`$+DG9AnTgUDrd=1L8L}4n(-m%P~DIGB6Y~)1QzW=Fw_*wcNYNTHV zJkLex`8K1&@qWr0k^b>o-sK?2TXkA_SM=wGJEqQc;j!N-?2Wt^6ID*ILLc?UtVa6w zv{Z|cJ1rbWYw0bDMr6`?+>Dwa&aU}4p-6m9lxmZfR;6eNc_9y<4`{?#D0&yPw}{cQ z%XERFg1|V0;c2#}LarblhGO3Sh_AzIt)Zp?UVY^H7~vuZmVP>n0*alQpIgFU8rPV95n#7C54pSp2I!9L=CcsCLNtR*k&6(5 zKtpF$xcUB>b~8z{a~MPm^OcgjHUd={ag8C#0dvimA4u@g47E+d-`=}+YRM3;t)1W4_ zpepJ|NJJonlQs*JAMc#aZm|0u3rLDDhwKCB=*`{)kna|Lcn3y5wX2HQ_{ePm=dxRS z<5;S{{RaQQMmp|*p#9-zs{8GD)sJ?`!}XMn)n%ASAahTUXCt!wiD$1S%SsiftG-dC_0c$Cs?Pa zQOf$<{YY>cf$A!R`{mgtQ+pxE6}A-YM#jqkN5(!GN_OrWJ8(g5#oh?8u#~D=Lg6ue z!8GVfIXHP&f$CyPbDswW6PLkBlv8}errRP-veCuH*v4 zv{nd%-Q!Dw9Ayw)^E@u6Q9XmkcH5G)kwJM2i5YMRPa3ubOzMkYZJZ0Lqu8lI$A4A<`C{uyojf!yv47O|F0I^#fbcsA%faXW|h9PVC7G8FI~7~TVfugNYUw(6|XfS zPsu!Uw?z(VI+ttbRDb+#GtYocr$7*f2LKN33@(brpgH~{U@>gs-QdT<<|Bb{m!I8| zNi)&T%4GmYN+=0`)07Flcm>UK4y5D#GLLPRsb2m-0^)jWt1q67Cq1O^1Y)X>N8UJf z0Wn}a3*OKKu0AY~%^{y?Qxm*)lKc8{a}Q}i-grq4l`*NtDySjKj}@6%^&hIj>e|fP^Sy$(+Y8h9p3Jj&p%$ug=F6(u zOS$94%e=u@?(n8ERpL5CA1lB7?ekej1qxq-uJ|&s=z+gDaY62(0x=>u*r+2KKre{A zfY=qz>yIf&_5vsyQFsdwy&MV#4`Uwc$5$nISwTxcriG@sN%7S-w|4%FW6URZ+O{@a zNYg&H=K9N)uiR@csKB4a+g=>fmK>NzMqPs$M(M0z-UW*L0OBXMv7F+Hz;o?9t)6#m8b zbd>S;)G*ayT|uqr6n&bg4+Ps zx}7GE1(b5V8jC*E^XJ!0a{n+QtD;JLg|U89W|SL&>*n}$D?eHR&DKlmG4C2W^mO+d3%6SVQl z(>rp!vrEHL>3<`&1UiR$gJu+|vkU>zMCt0EdVO|3@_l&xG7t8W?LKe=nRu?WQ`a2~ zfaw=>RoX`Pyx+<0hs_}NHD31CyyO?xhL-sreE0lj#vYHnW-fASZ{gG9WlU-Ej>i*$EREy>0AO@1Qc+UDwhP59K{Kxx1rS@S_!iHF@34PPBWsXb;yx=#;^mJ{>7G*>UsQP1Xm6QL zuaAuXW*Jm^iM+e(hjBv07h!h@i`(^K$$@xuSbeI%pN73sh>Gg4)gA>J=s>4dCjgR` UME$MAa)`+?nPrR}y(Itu0FhR2p#T5? literal 0 HcmV?d00001 diff --git a/2024/06/10/startup-data-health-starts-with-healthy-event-tracking/kafka-inspired-startup-data-pipelines_hu153a73840b0658eeafb17cde694e5f88_121924_480x0_resize_q75_h2_box_2.webp b/2024/06/10/startup-data-health-starts-with-healthy-event-tracking/kafka-inspired-startup-data-pipelines_hu153a73840b0658eeafb17cde694e5f88_121924_480x0_resize_q75_h2_box_2.webp new file mode 100644 index 0000000000000000000000000000000000000000..32bdc199863cd7267ac49f0153c334667644b490 GIT binary patch literal 21714 zcmV(xKQ~&^1MM6+kP&gpIQ~&@lF9DqaD&PV906wKcnnDMzFH ziFp_Nj|cqIeMk6@M*M&F<*HYWdC~s&_G`-@n?LRPj{B?r4~*~Fj~NS#=sWLE-+yMk zTt8dSALPA3Kg#*(|D(!B?%(!3Pe09i$A7)%$M%2toERC+2wTEMrqN7lvt1-?c^_a1 zn(ahuKzPU~Kq@)T2+4V;m{02}PV-1)DYL6tuLxIkrGJ>zC2?#(o&v zN=`fq_~`K6|5OZ-<{mZ3aCjLU5XOwE7RV;qRc!vi4+z;QS^hvPF)8tR!M)-*76OlT z8Yq*;3(+Gjb(B_OilkenlBUT@6rC?--~7FBWYMi&8X32=T0WX-2nZnzD}mD($n`9! zb}Zfs>4cf*tSLPvJCcRj4?1<6><4Qx@UWaN9p>(h{>2a*J`8Lntgrs zGqBnQZ$Joz`-KVNKRJDJ87-GHjNFSVs@`{1ar0Dq`@Pgrv!Dbk7RRhm@@xlPwC(S> zBxN){L2Pt;NKOxb^}@Oqh{?y7GJT(fW0?~9#fB7b?%?=z`s0w0Goq@i$EaMx7`9a- zR~~woBD(cqvM22uUhwQPosWCZkg102+B+v!?i3?-e-mqz8X?K#9z3xdD>mCy)!*;Byxd*{w!rB_e~M8E218Kz#`^gD&Pd(mmffbF$oZuHU*_LZz?<-s=#%Iy80vL!=v~TMXES#1(0KH_2a)6Fh z6zaR`ic!FHM!9{&0RRiv;<7jL6QFTxfoagO1%McpUTfwtVj#A8*H=U3 zV8V3K-d-iNO7I+O7>8u`D{p8ZtUh7nX<;UFR-C+zw|<)DKngZ;`5g;;GbSInJ6{^; zo#m{aUQ_^S$+D-SV&;0axi_g<=r+li=(v zoRPTLAKKKcjW1!!RjqzfE!KkxTss>1;sz0u)0|^i_xd_e*`=xLfcG;0Jufu%U9VA$(91NiyZbPF*gz{F%3u5p(1$_ngmD&?yQ zoI$!7;fj81q=(0j?RvnG1s<4Uj!v9v6Fz!6hLGQPEO{V2+XD&N zu|N1~<2|l^gP>d{KFQ<{?J3kA-f;_Nz%M&qX@7y&EnSGLwwzBYSMeGW`z_KKb=Cv| zRW%XGjWEpeHy!;F%ZnI;1vI2ay(n+TVR@KJMQ7)N)14heI=fZlm)8Ux%Fwh>rz{a8 z5$&i8kFoKj1FuH2mhb?1XWXal9R@%B3U!NpvD-UymkZ#!TVjO|EuS$@h=g&;Gmayb zGz$jyMd{ro^pqx`1h&2%w5=Ib-x%ks94kz*h@^ZUGOLc6fYqCBXvGYjIq}RaNUfdd zK9hD^vJ0EY#J1l-3mp!Bzk>kXS1xT==oi?C-cVaj2PibAZ$u6%AIY-#e$CAh#&EaC z<%J;FkkjDLqnzCD^|(j#Zv_?CJtYP(z(rkr+A3 zt4;!m?NrED<=zgBC}%Gx=?_LpE#?J6(U!MIBzs6&(&O&a?V9X>udlWE8x`n1bJ5X* z4h;IgOb?Sj#eZ{jbM_bfpV>p#=>W)-hff2Yx=gzjf=KI~2K@~Uq#~TS&wmuyxbpq;*Jnypy9D?a~qW$WkD4Sph_g{wwe2yw)2^05A z*-pB^??Yjr)1URQFTaQ=*fr?~A!8gKIcmQvaC2(QiajSpjH%SJa$NH|o(HX6y@*(5 z^gr>z<{g81vszDQjN=CyRVZcQNNwXo6xBMC;N!ags`B2ms0X^QgZn?V!kgH-*=w;_ z4|bt2dI04PV}JL^8xHyL;l@wz%B3uv*;wUYQ)4X zOcBJPGuh*~B>@Y*3}E1BjYiJP{DOVXRLnaLF_|daV&PVpvNKU9mtkFZ*|Wp~x8^xu zdQUK>pR9JSM)|^T^zTw=?O|L~Eq?G#WGh()r>0S-xIJ+2>aci9?6;^>-!*(u6$+;Q zv-N%D&;LD_+9&^a56aRz?Cho-hC@F*)j;tW`q2(i5{jO6$?)3@!i+??vQ9)s8{kf{ zEi%wtw%pVPzn=VP*%fq)n~T4kp*77S6{9fS7}y*WL^B~58~mrs<^ScmKmJJD(VU2F zNH5;BK8$q=+1l>$GmIrIJpv>=__!qiQ( zu3*hP-Mpk4tnO)-+iAG*G#^A{sR80?W_Kt$CBWH>HVWXz2}nr0dnu^kdVVRbiu zSi(nsWa6Tp22V)iQ*caufB^p46~H(tfBp>+N*pGDS)S>n)fU{C&N^TVWHZ@;W-Vst zUp#o!xIx}?-dvCV|%#P~m`Er_f@c5DgkTd=)VHSHoDVyH8(&<<(!`bgWZA zVWwLc;9pOj1_gYa-eP!?5&=rgZT2hn)YvBN*5m+!=b581J{^ERPR;XSVFg|5MnL|a zX9&%7Is2{t;{H*O0*?2(EXgx&=ezX&{L4-PAaFMtwEVzj^EL3U;vUUfTHmUF;Cpyj zLQjKf*^ApXIw&Quo_giEnQM(Y#iOa4I@^&$c_GN0-dYBjVr8JPEZt}fDdOOD3qDB( zUqygJ3t{cN`*UARK|t&YQrVTv8J6)LWJ)zRrZVHH2@ZW2;7roAxO_j>w9`OAb_G82N`ig)~taAK704{b7QV!RMLq+=g zU73?-%R2Bn-P4Bp2&*)ZBGXD_Vd6P#B4U;5`}3-O4Gd%{`T9=~_c<(BYDUJ3i^Sd8 zwSVID8P<<+JSsTod~adD^4)F+>oD!wgMn6zkN>^g`3aMTnz?Z6Z0hAP&_^6qgmmyn zPcCQs3M5Zx*8%`waK3rq#Lb&ofei6iLb*+yr&kVxtX?1uQS4aU6=|G08`!bmpQ??G$uKRqFh@c>E_DAG1aDQeZr>?TNOwf zXvd&J!u2#a?Ilau#UBX6zMA98Uu~Utgr7S_7Qu9mOj-Jp_4WS#jQN-RHISr^)B1V_ zncyggN~|78U(LXP?1o}!-=sJ$z+L6*I@_vg<%epG*72ovy3V!gp^w+$4tiUF0EylR z&yogq@P(*5*wzIp%H#yKcy9l;D2>4z4N|DGA81yDyZ0eEuUL8NIKI)i3dPP6!K~Bn zI8gToyWFWqI87>3h>8}9Yc7Ph#YwC`!xHjPWZT+TgEl_{yvv^CnZgh3Qo+W5qWKJi zS%B_EH!fGGLJ!~#J@W=*RX+M}zkZlKR~@#Ew7?RmN;7d(nCtlzkJ(<`VvF%RUvxXj zP^YeMcK?q3wVOJo+vFh)-oJ@tE#a!EJjboDHj*c%t<7x674LBrz<(Ej5N13n6;-?zde1*V6W3e{}#MEkG4S)>~CtcCYF+lI3YN?xTo#m2*_gYAxN)80ez2{=|2T60C zTiS*;_iO!6ZCdjCB%nl_Gi+#Wv}y$yQgNg@%gzrKWZ~}t_aTplVRyXeRds%qvNeby z{W3hLDTlSV!iTTIfFqgXsig~&5x#^&pH&h}c(a+%EW@O4U~ROT(Ml=U!O=uM~U@Jv8fV_W%>m zS-{Hs^b!XOuZnV!t*jgUR@)Oy`n-mF?1N(^fEk`!=z#JsbJhug!57Ia}uymdWPlCyqo(RPh^Z46{_>@Ms1mnmh>EwI=&wFO0%=0!EP}q|f z!|&Q8YThJxH5;<>#Fx*#Tm0_}g&7bWct9JaYX#sVQV3*7FFd!5{2<>Fq0(g%BHg5B zo&q`kVZUZJTUU^H9vt$60h?=~G${Q2Cp1z`+8w6QQ&@al8To&Xtddod3S;#mE6))l z`8J@qW8@qGc_PnM2{1?$__6&AbmXN(TI;>{v&+a|uv69Mc(#&yvpvg@Y4enzQT}|U-ZEK671nSAoZ{;FS#nbEs(Hml z#Vfy=lB5}r!wb9CEzVhX(=B{(av&8e*}2O8(DwgPFr{LYt_#Ec@RI&J>bgP+n-ZMim;^Do)O zudcK3|7dzu#$||Llm(@N#EOv_xU-&0_~8y}oPkN)to*eTY7TuJ}~h+ z=*}#@R~EFt=xp#X?i01Q?^m-MD>D_y-b*k&6|eM-L?%pdKf4)anxvB~MpUMwnBnXd(7$D3xJm6Kf2}VkwXFb5P1KQqOFeq((t( zfQz>%To-yLn%0I`=Y>h&UhNxk*rEG&^OuhJkjs}a1L71t#3CQhPEl|F^_keKU!s9w z?DNn~m;IX>MTV>n6U%Gu8NsB45ui#R-R=g06w+yF@cY0nB_JEh&6Bq2wJ_7E@;JID zzT3oD<@bCmLP7{S3Ht>r;>=oAhe@6hWmo(FKK7%OkANzKb5qLGx4Y^|k_H`uQKM4Z zpikw@G2R@yqbmpnL07-djb7-b^rAX(S*TlOuWZX|>*IF6A(o6A38k3xRa5u6MbxmD z3qV<&95Lmo`mxfTdg(vGnrP{-woBtEwn`!2j1}!adIg!{4V1ne@muD7zMlz`6NR9= zrl|lDAB4|grHY;7bZ*bOQRp8g@4IX@Bf7BG0NP);UJdy5=_FA5R5~o(UvwE*P$4Fr z8H>E$fPn4~U|s0E-hS~_E~q_#>ex~I|6ER$qx_xtE1DnJh;vMlqQI3FJ7ipnwIC>d!vQBqq(S^ETG;^3*+mw$8NOutB%n45TYk79uU@(#`J1DX+ z)>=wa@n*TsTGc~~PQx$XRe;rqqe;N`&eK@;{Fal=@NHrhHe`n{MDFs&|B^i)k0vIE z+hBPoQ9>(HchCXhQ~Q*hRl`91lhP*rIeoEN(}IQ=NDEDgv8om=f6RyXzon-2|Gj8_ zK5Tn^OU;@P3tQcRD)KQl-Mdc?eAk!O5sTF9m{uQk1%Z3$o{TkS_SSLNhbk8Do_h3CjNqU9Jp7P+{TG=ScJ+IdB*(tdd~u%>sj7Y@)37AcCjQ+mAdq`4tO*W09XYyP0aX1OAFoTsIV2WeD#?9(;gGxS&23d@5o z)en2*`Pw+KV;5}V`-d`{ciN3J;D-DJO1J zpSqh(O3ory)=%Sjd|=S=pfG!4NRWNS=`(sM{gxdC3r8vKv#orS2Ka)Xj6)x^k(72E zEr6$XAK!W&^RVCynvKrUv=t|nx0RSqT3zg)CJEhdxuI^UBA6{k`rR&BltRODPXV4XDsQ)h z7{-Bdvg8jvTyr#1n`vM?zDC7ZoI?`vdANE2_wkMI7@){XUgAVwqR4`QjIHUgm?hB^ zV|Eq(hB@XA{(F8d9Muct1RYC-2{XF+bf#|~?QV9L?mkJpHIdhFdgvt=Radpt@9BL{ zEo-oIZ~A4gN9};%C}T$V=i7m{$_4WtxkXd4!;YqrlwGQ$Sn9&ysahNu;IbFcZ2qfj=>FbO#<&sr`+HlXkt0ON>F$u`eU49HoDCoWDpp9ys?f;eyOb1~GZUMP9 zqm2lWnMQ+kEW0`vFBT3&t}U>omFb-Visy6UZh$)^L=bnYzdMaQmcXSZBc=$kHkfes z=xN|+!H5$y&^V(XG8>IHleo>XAxDJX%-ZVezIvF>pRwAco0a65kIPGzgKHcYTF?gY z4iI_tzb_94(pEF~?cg>cL(l$zo*cVfv){H$7DEI%5DKEAtTalta159Pa*+JKP6|pW zvuOjQnE`ioRqg%aH&ysG-c%aZ`S$pk3bORvxRyb*I+*iHcPFbLv+ES3$N7INQdTX$ zi~6GIPwd0=jKG&&W!}%|6BBeTyj`AV>(j-KT{VP=8+`6bGVQ5pV$cCFT<@# zFu#6y!`k+JryC*u9kFgbz2|0D9BP=zXzmaY1L1%q1QV)AmQ&GLf1LRa@w>Xy5WYm&foJ|XEjCQW)HA{w@R_VVrK?_MOCRkuQa|1jvkhU_ z-Dw?9%Mwc4B#hgf><0U9cL5e>Pr@c}5Rwh( zMmkQdbixgvRr=gGA0Kjd$pFp>TOY5-0y#!@@&_&W38^lEJ4J81;KVDOQR2fh9Bw55 zi$4}mQK?i=j=L%zY(yiFXXIFHRggj}N;zfC&UjM=Wk-^@JaDW)uY=IT_-$#kP#4H+ zKxtb{HLUyu5*XZr|1173k*f-4H;q73$9^qK$n1lNsU5!VytxVkV*aIUkI0`@0TUh^e}y8Og@mef3_9L)BxiI>pEne?zZd+PkeEW?L3IoU%4 zT6U(43w?O`lwHtZ1lj6C;}N;lr--}z>^o|HDM;U zk_546*@vhKir9u=^ZZ1Cj$-+E{l9^e;Y4#U+qaT>eQ{5@96NZqnVQ{~F#=CBErURm z&V8o29=K*zan6L$&xjENruo@tiN4qa5)Id8h2wsLphXqf4s;y0)JmF>#QscaDpLLn zQ{R2`B!B=lEKKd|dNT%c961pIq2|aR-=A_h>J*hI2OIj^nuIR6=EGOXHucQavs9(X zf%d0NW-Yakk0HK|oruP$!6xCY&Lj~Ff;pcmGjEaC@?*2Y(J(D>0yq^F*1O+wQkW?_ z0{NISxn!|jP7X%*q^)s#ofOVZ34r88o<$J*A2Oy0>fg$xZzB&?h_FcGGwj3(xucUc zN8*4JS7Z$jek?IC(h%u8Jgt_ZmdLVIfTgS?gTcS{KUdd-Xb|NbppbHx63#YJQ`1q) zCzI&56u`WV%#>sIuz^0o&d|yp9t!~`qdscwWP@G7XK1j9^+wgHN~O*|1=TLBf~C%4 z2Bs}e58pCU-rD@_lNz($u#hKIJl00*vbdAO)O4+tlQUeQQl{iQ$@|A#z#8UH`f~Vd zFoM5*R30tSJSUjF6>Eh2E#Ou!s$pdziE$D%;8TP17sYN7}Q{Rt*#uIFbDJUU_icp-aw`S3{4?b-x6gUHEr!t z^8ow{&D5Ep9yPHVAB#2#TB!2eqlV(EhosLtaiSlx_Ul0QA_graWJt-Ni1=uk$u@iX zO`f_dwftu3GL1`O$g1mChSd0LWh<#Oq04gLSFjgyn#9~yVT{C{nUgk6_PQot_#Il7 zPN~qeUdyM9RhPFH_f1T5EZiuJb8EG~s3d({jNlQWj6XhYrdn%xj?oj2#!s@V`^($| zg7V#&={x#*GP5Y|i*oPDkwZ1ISl@LViW-5z=b0jsI6QIx5#fP7N3OmnP3DcY2AqPS zrqH`SP^v*rFY3wn42{@hsHDr1)Pn2c0XnI|NG)eDDg(iV0PPv3HM-mc)N|}TD2#~ti=~(|IcmVbtizq2QuSO&k$eZ=bhacaz-f@&4tip01axe+= z%EUlIR2+EPH9cbN2%#cfQ}p~S6KU73@Ybm_woK&>70EYTw$$)&eC=K}WZ zT8Zw#eT30HquC@rBdz|p)_Qv&ypFJer^^9qZe8mEX#=I&wk`yH*$$nVt_^z|wtW`o z%;GBil*Jv@qJJlCI&DNrlL8^e|DQTK_G_88^+8JnGji2Gem?AQ3d{$6nOLMMpG7G4 zdEmH{s!R^?y+*HZ^A_Ba%x(1L#4viSE%^N)4T2)59C2sHl)ibkIo+~DY>3QnJn+rl zDAn^+&Kn4sq|J0|6u`tDaa$*YBhJccx-S~>>7g#RRdbP$!c|^W7gby{wfFZB$JRwqwAK|uVB;M7Sg!e6R!j-U z|L*0=9j;ec<`0+==}ZjNA^b6<3s~+_T9zNcmgT)iqW($auiwHIc%q~_OB^UM{XbYC z`yMP?owF1$3&22h!^>XN$>TXg#X=8Uj1C;AA2+ld7_X9)g-K90gSS93eyPdnJOd}x4id8f?g ztH5)QH^a}eX=U_gVo)}R32F8s(UB4v2YQMPI9ZOC&$RD~@|0OaaYJrC-MS~N=K1~& zbLz4Ym5g?fye1v-G`_M&=i%@!nsMVyLzO04H-6JUbPnF!+jYoR2VWFNFQDa@^MQk7 z@O?{F#GU6XX6TC$pqGlgGbZ*5@~-cxU;+LN4h;IE^OJ0umr>V5F4}|FvIxrJ`;fw; z`*gOiSQa7{&z>-s`50e6BBcfFOJ;$XscV`q5%-?#0Sg>iqYbrGMq%`;{*HLy!+nYT#8apMVS)F+4azfEa2L$2lN}l^JcYtE_M$ZQ z0Bs#Ii#C$g`WB}`<$5)9znI?ooQ#~i)8|DC3p+LXxN-k`?Bo;88Tac!aL*D2>weP4 zMysHkC@O^hgThr`$G&N$)fkUqhki_GC3%O}B~|c-vs-_}TdAmd9!`pD1)E|6b?4tm zb7xt7dgZTGWd@xlJ7*kF&U_Ew!*~aYWdiE`I#TWq7GJriaaES1`2uhIQ;p9vLtGRV zwKlw*^zuIHv(Na8h@8+@YTfQue&Q}{qsf7?hh>a!&0Asv$K)k5{gOPDuF^s1u=bT;UArg{Gpo!D6pG`jVkz+AYtl1<@?#rp0sLONuU_| z0Dy3e>=Mc1jIbd@$r&B*?Ebxfz{?p-d!c8<)IA-WSxPR4=&}a2NpG5}#AENpG&B_b zbFaRByQ-+pp+(xVIhDSAPcOq{0k5qKm_0W=K#ktb`@f}`Rul2!UgkHFRa;8s_u1Qja|?j^fp?M?w5Gi1Elc_gFy$Vh@?q0q4n+*U!= zIEQg4JqYm@V+Bh zoo0jmqt+Jbn}Yef-f~-kjIKAgi?%%3aC&Uk@RL3tG3pOqZG|5;oTB`^T7>rR*xqJ- zgoU8zOb7XN=hu-%wX$i!ppbmBpw-!H#`hhXij+{Y5#Xpb`I=zg;vCuNqHDkIXxSnf zJH_!}1i97Ind!z&Ro5MK&jACQI)$93F-ihv+zxrS=VXwap;w&^>2Dm%KE?gL&r7`7 zTmeB*V(#}P+JFW+-RkmWDAS5!*Y2VhNcRFUIL^9EjK0fopnK`-oC#Pu;BHk19}hyx z+vtwrfP+{ubQ5n`_FXc1>Vpv(mlIcW{N_v`NhV3`(LSi&wZ+tA8Z;7kJzM-uwqT&n znW6V)v{2+!qPmCIV03hu-mL;nVyPdk;)VAnYGmOK&9+&iAp6ZsZMfT!TYb{!vHgh(~~pSD^5J)GB|}W8%=VuD?l1-DNV4jMb%`0q=G{ICj@>M;TJ_ z6569_pkz0S3_2^K0P+M-LP=xZqHEnii~L{iK?uniFc(OS$5)$agHN;4_RFj@;@y|! z_3rX72~lB}>|sx%!oL`T0o%_?19x@nD06FE6v&VKx0LK?*0gV+h&4b)?4 zPbHQQ!r2%8OnQ=YGDYDFhE;AOYgq*PWX$R0W88|puruU6E=332`hEABTh=*G_RHl< zPs4TOMitdt!F$SEfsYZm7}&zR&0O@B6jdqK0Bkrsg>rKU&GCSRXRw1=$hujoL^TS6 zcN2l!bTVwcqmW>SHbQR%>DF1F9#sKF!)bw72akyC<5WZM(3xBR`McI%nyi`V`qi|; zOVCW0hg5Vvat#-IGqw&zWH>QM^mtDqnzDCMJJVi7U5|oOM8d5{1^85JT%I)c@a{du zke2uz7#tFB_`#+7Ke&}zl|&OYqC!XMY#5q5%$@M=qrf+WDd|0c232eS{9fx27;r7} zU!|fi90S|?)#l2nyg$R_YD6h>gp85Rw!Nk^hWR0-)$~V08N3Q{mOQ+yYZreehI(uW z53jB++@$LLFbrT+3Q6}3h;g9e5+y=dIKI?_z)&%CV=zNssbmSVq7V1%b~ zDbL>SmLkj4v&2U-F~?vlvBnkKlIsux`fjT%1%y=$6 z=r#qaXK;@)kNKU3pkcblv=inm0-ckVDZ3XrNsM3c@?rQ{)Y@1o07|J zp8LT#prV9|V$*$6IGggtp$g|Pi$R77=1@$o=BUM!*(m>;%z_hxeN-)sC7o7m9co^D zw_*l5FUmSn-CwIr)3CSDla2UCEhG2Z(a_wC`L`bi6UoSBw4l)^C)3yn#Tj zD_k8jIj+Wn`Cz5?;Q+p7LW83zt59mA?Y<)uw?k0|Vn(DF2bW?1Kr>M4`hQ?`d@*W000>V{ zd%bvP2!`Ck3k*Q@tzIE+F{4WYZu0nrMR3=?O*Gy6&?Z$&;+7w|%l^_^Z>~6$sxT%S zj8X}??V~m)b2$dfSVqz5qmV1=j@gF#h4Ai+3}8n!ZnIYvA6Hxo?@~oE!da^mFh*qC zoahzw3~>4g@+$e|+QXj**-HkCdv34vct;T~hmCh!)r+XV$XOej4cH6`SUJ$;FC?qV z0#9$jYkLu_p0A}MZPH``9tLU_dr@!@MCMI|wjP~yTbT|BFk+Co>u?x*qr4@j$XpD% zl+#cgvhbQk`8ZUZ%tQ&B>a!g)?sK&9LO9ut1;CW$08BzoHes1)5&&gdy(oG@1FthR z?>U{W1tPmR-&=JV%eY%gd;V9dCVpG8x`zDhmRDhdJL~ujokRTj(OS1O3UgBYU4{k^J2s&80izDWE$I<(B2bhB3+iyjYqI?g z1z_Gh=BVnB{B%Wg9iohTY}ZnIEwaHIZS2e3{y(bcmS9a++Tv)RJi7`U)nCLZi|yG0yrEY_Nizd zn11VGU$kNOJN}~Mi*8nL8%G@ZWEAt%_3XxYE89hJh-hX4Fv$XRx5h8!0K301PcO-o zl+m}!8^n)n8nUIE36i0Di3_D3O7tt9^Yki|Nh@6G<88XrC*luom0PLxI>x4R9=d%t zdaCX3-!+cV5J&F^O(QeByBRU`Sf>r5vpCp72x{t41Ormno&k z*SBsU)T5$Uz<V(uWi27g|_8)Q?89^&)&bU7Pa^|sD zLZ(8mKr7?Xzl;W%(7^8t{%9TDN(IpO(B}X~h`Maco!XS?q{-a@nTyRgj>fno3!-23 zWq`C>Z4wRKh>|EGDA@1Tp3Pl4&H?$H;nD6pTM%%8m?z}e7EM|zQJ!n~KMK`T5qrB4x4k+*i;%B(>pFe9#c9M@R3Lml4E%z2?f+2&oPg~1vJ))GV7Joi*z&U?bCMY8&6=A0cL^S|j-?nfZZ8P1M-RKsVzrCmRDVOpw~Ke+$@u$**dqW|wIP`DZS~y# zVGbi1=UO>rkKv9dCyVoi*uR5T=cGW?(iEqWGXwXD>u=IC(h4$?>TcAmL+=(?b_V4nNee&Vj{vOFuHq*$*&_FJ|v&@g7i(Q)JOS#9zIT^>` zpH0?vum(BffZJl}2Q1vM?-Sdwji!R*S`l`!26*-F@b=RN>E+i)U^sd*EZvYcZNFJM zXw8TQVP|HX_c>po9by~gJ4H_v3cLlVAUY*hg|`9L#3>$d><00!kbCNY@w$gioI1@i z-sxds6TqLeLQ)p8J3wdT~GY`ul6ss3mnxSEUWrGfN3ZOl)Ud#i4ePsro;l zOxyt8?=Nlx2U*dLN;Mppu4XE>nmT1v-A5jUsgS^s#jZfL!9-(YYZ$nB)-iq$ARKv? zG}k@km%~08xAh~Mh8{jnL_t{K8SnItU;h#rhZrT^VoOKv6-4t$BR;0z9?!7u#G8ol z63$J%Z3*GT+#JnLArA1HZpRrwsxdeQYoI>M(M(22EhMz#8rdwABL}7k1l&lz#%*mi zqod0oA#CY}tI2KZ2;VSat{YxNM}gD)Mperrilznn{5~=hOJm zTB6sog^#}&0$OpMW~BIIvFx0RiRDl-aR&U?Qjfy&P$*9PE%JP&5*a8A!46~e+_lw- zoarzS%1Qi0FCIou2v-UDlUHAmC^z5oPV-N9h_(Qd-o69wV5Z7a2o`zAonCv!@`*aV zOsbXi4?}F5)J$b^2%;z4$hIp%q-ZEnuH4}D8-sLRav3x0_EawCB^NsqTFtNTkToX6 z@<(tOEo~;o9b%^NEH#6E&ug6#c)!t*22PjI^lrsLI;d%(TbuRKGtCpMcIuR6-rr*{IBPek4fScc=YIf#j?yCO2aQ+?uQ2!YYMV%_;znL8-S07FoY(Bfrq<&|X?&7W_> zps0{o(y3!vj9Ks(8h(79XvZK+eBL2&PLuju4!X_%ZtSq_YtdjU&4<&rB&G@A7f6yk zAoFA{83EHQCW_0_K*a6%GFWa(S3*bX zXEJ{-Gmg=gXCL`Hm2urJhq@x}hY~$hqRM9LA$~nC1NeD?#Mmec#&!Fjd>0EwEae)) zBYl<&Ds!jpFhu}1sZJ#;+E>TuzEG%Rv2ULTazkcYrqZYs5XBT z1}e*xu+|y=k*QV;78h~h8u^e-eQ-Rvn2CE+Aq~VU;z?DgS?_3Fy%bf2t~X-v zA}cC0W~ox5NFXqRlVHEh7Xq3I<8lele^r!Jy27Em8mETgokHt*-K&;C>AqN~4ZlqmS6Gl&LM2IX5Gn=0G4wC{+sZ z;7jE8+$C!|V?gts2{iw$57U%4*s{w*&c7g&22b%xVv^S7Vp&Ek#7RY7rq+MZg4n>v z1sCOu787V<6+gOT_DkLj`X7K^b^C`cuz^~Sp`Vl~D+%yZi)*|bdD+gg44^jVeqD$s zw8mN{x)E-$EfMGi!_t96ywIC3cFQHGOMkA$%U9o5$@Zg{yIDY(_haqg*CL%?%{b(F$m*4znHfJzu43P3j(9`xW~_)9A+&7z>DqUG6=wD3cfmi zLMBfVfBBEp4?izgybVt2XaOymJ2@i=kiYVEY&Hr%RT55vH}6;um5Xfc6_dq%9jU1W z4!Rct*)K$-Js+tcwWifs^PcEsCuBlk$AXiAoEEd4z{sI-Psb_8xyzFT5e*pzMpP>V z3|jxQr}#gxavj@QQ4YGJw?Mp}fH3tHd^1>}&3@ZP&5#sS;jt?Ww0wE8L=5iAGrBHEn+<1!Ovx@n{~1{ zf-TEn+yrL7!1FB2*c$art+9UjSvJyztb4B%XL=NVuWT`j-<{v_#y9^#$^1j)uhjwI zN;WeRi;JmBrzc0puN|1mpn{9p=_dXXAH^yWdU%MS?XSd?SGX~q880KYe*7&1O4!kk zsz8Teo5Lfq!l(LxHC%-4@8^BB0JJDUZ}<(KsL_OwF&$EA9u^EFMu z9{k~AfA4+>pdLhTeVWxecPPkD#dk1u1F&cvUbrs*twsp1c+kTUWC;Yim02kcNgbUn zXL%Znsc+EOj#h(0i7~3P$rYG=7X^vmXtjGQpJ$?KeK@N^>Fj;@=vGKrd2q72>935J z!yRL-I~ar`V?aR#07D}EI%^RF-7>kdjj8Q9c^I$_eqUC3tZ`M)7XtaSc=jO6M2S0m z6|QM1zl^>_n5;HbzPA9m$ca8lDLK5bFVZ7~%HNqsu2(K=;iKCTR$_A64OPrpV0UEW z!$tLqWb2Dm^@IB+q5=Z18lv@=-GmI@iN*%!b+=GqX%2}s(_a+mKr!b&>zO_sW-czZAF!-;>c$lTTwKsbM zAZEs3Au7&zHy>_Z$6&Z~#}b5R69FDqodXBcsVr6?FSNR*Y%h^|~wXY;2v zwo_&zMc@$TXN6dGdkB>kl)3Ffs!npD$g|7&zBaa%CV(ftpPIvg&YF1(ww)5r@V#NZ zM2|h_pSu|5Kl|W9o_UMuT=}RFQNC?&ph_HwfkjSYH5QNDTSHA zue?&rv;`yb>ZWuDv(}xsC1gD*k_V#PBvcz(8|A3&^oFRtdpY;>o`wqkO6fnT2BJL^7T zA(Y={Dkw1y5e}YNl!MOlZyvZlO!5!>hL_fEkrOj@&!Cr2XBwZXka$f`!UJnlhl)fN9j;g#jP2vH&6h!NLIB;vCM?$BXfv(@x z@viDrQfX@YsNaLI{LJVBQ3Mmzya{^l4k|$$EFfT50jkd*J$*eE3Gd(zQ%b6U5m?Dq z(idO;9Wu?%@iGpYdarCrSC>PSV2+~kl-2s!8F_bb0iempbg7fY+i;O)L>f3mA_QWZ zkblaiJmQb(KEX{{TOzDLzc3Ot2+d&?#@ft`IPd8pc2i|g`emzy!j8}|Gfc?k{nd;K zfpoh=We-jBGlQKi~~R)TLMqLBtYiBhm^>ZpV1G_52T6Sna;CfqmlU3)F{2R8v;q!a}e)U{dG6zwL3e($P6KC31*88`%;eBW2j)Op)q%Wzr9qArEba@w->_ zOY#=hN{?B9b>i@{3I_IcLQpIE|FrFh;{BXDy4cFA6yTh6my2^l(>t(-Dt^d68iFUp z!_x9luMk2fIA$#Eg+Y46mdB?vhcrZe?@WDYqB)vzwhO z`SWi$(w?OLg8qW37I48HEP}N8m#h1oos1CFav?^wR0asIU~SVTPdh{GManh{&=}?) zI(y_vUPoi^BjPXRSPiBOe#yUrJqY1}LGgbUudw$19H-A~=WNL9%H0uY{T%rudZ(Ki zPXO+c|3HRn+tnFu(ekAUtVIR9XL*y`SJSPH6HMFx4(nz?63(tioLjK3kEqYY{4>m6HC2;@agBrzs81tBuk#WG)Nev}82 zxC0+I^$t}n9W|li2D0Y&>VETz4NC1GKtIdbDo>W_M+|)N>_c9IXRptki4b`>tbJsZ zz^o1M@1dBb(O8l#sG&68X4xL1}@%*aJRehOXv zXnVF#3paa&#uF;w6pA$y*j;E!>xnT{U6K zQuswYdRP#{VCGFnXyv$7+>8A0tj;{j^^EK&X9A1X!y%+Sm*lx1@S(6cTUgw^c)zw} zYQBpY)j{NG8a^syO}2%32Z$q1VA_kI{JnG>N!yxzrlK;@m+B)Y(c4-C36gR{TEpdv z*tVuio9B~`ha)ZrU!ZYtXW!lz$kCwJdBJR$1)+Z|2SMz(Xu&0iYPsHC{VQjMv0Yh-JHHTvNw8mgkJ{Y+;tt8;R<&b?ObZt~2a;dxOFg~pY z&Gq#D7dM=5bw;27hB2g+{wRK_-s&%t|xkc!=m=n^#Wie3EXI5@5Ys48?RUc7N-h%&~k`e`+XQxD$JEFT)@|#b50^EO=le z^n+Q3e`MW^jD<1?V}b;SBU94zx7(ICVss6_+Ku4KtpyEYL7DNea3o7t6 zCvO6E#k%dT@|emu-%LdGmo*1Opui-q7Dpv?^M$DveRtc>*urFTc2Q6aUX_Lzd~8;( zsiyajliT*bS8|k&E5SQ}^1{1$d;(^0t(i=}Z>#^#+=ZN#Jh=jFoW|7v^1nl;)%8y> zg}76pssmwfAJg62&kTpY*=dspudAgm*rtA+J?kQ|;)`l8qZ$d)f*yF2FRj6NOXhIl zWW10K<^1p788o}sT^6xKaGvFlUHD-xQ^YR4eIQXEMprHTl=kP;2=!D0%Ssv6T~>~y zagt*mDP{C>eQ9TJn)Eg?wrDFd{f-{0flli1sPpPF7beZ(Gu$jcy>_{Q0^bb=IhpGp z{yl)upD-DGK3#eDThe0gLx0F&ZO}lQJMVvS1>9%)DX)L&kNQ4q=D$zsYa=q;oJ}>} z{(!un1Wbu#NfYE5;8f<{!XG*&9ysIy3MSjpPB`Fx7~T*Q7YJK}EDu;Y|5YnNnXUT% zqEVK+E+IO&WvwPT;NA7UK5X_*c)v>bk9#ighx6C%lbmP~U zSrruLVMdJfD+tKVK4VL=CAC|v95@KVi%ay_0$j;PLmEN#A}!2`PkO*xkQ9=)HUKFt zA=ajsYE8)Ix<&FD2?z92we`i?N|JlriNbOnw`2__>4ZHysK8CrmUy;vZm}0n zAe@h~ST2@4Vp=_Oi||70<}>p}N=EknqWW~f^_{qa8VKb>)IQ%|niE^%=nNk_KpT}a zyuJR1_g`h$(z*C}X0Y$MTU}?``8AgzXz3N8%$!XuQSG_;E%)5N_u`o~mg5cAjAC)o zUn4FK=UcQu$AGzhEZvF%0}K`FFQ)@8v$r5S5-oaoE)^z8ZkJ}JT=NxTi%)`oh7avx z2V(1a%>@z%_*L4}&};2GIsC#)Bo}2sY|X9C|5fo=FLY;tK7uLzTI*Svy7J70Hj5&d zjL~wlwfc!Mygw$$g;Li>8I3b|Md7gu?g{e+6r853ih;k}I)!84mn!dhx!qsOhVr3h z323!~$GM@#1&D!xy3Q-i;FmS(>~e|=o#!k4c*FBfwVwCX(;ztIh$WubQO}yM>-Au5sV9(TI8HK$u#$oQi-__1N2>PHW%<-#9%<9M~AQ<$j>}$tHE*ZG_fx!HwrK6`OdA zDZ;QIJSxPaM(zgY4j}#*DDW4*YwF$%WB2%Uq0g%g8}svHdFHBE1HuS)DD&Mas@nWC z?Z;|zT`P4C;D$p|ND#pS22E_5$Vc4I_U0nM%e|NbO= zrbQci-W=47-&b^DHXk53aTy!*88l@jv)>XC6J0m_SO6{sUhI*H$dpckkin_-iPe)f zP>1cxfjkh6ZH0!6oDJ=c!7K{YGdJ7q!KFBzBrI${v`{7TIiV`ls#g`ah=$3*{naz~j9#8zB4OIl3Bwc(m0$H*vPYRRte#<6W} z!A)xo?xHF(Bh1QM1t)MxHub7G$cmrb1s#`Meql=u$Nc+;3DO<@w*nuei41EiCPusl*3rZ8_USm$%0#|jzgzd# zX=+j|xSa+3Hq3eiCQb6ir3jcyg=)B2uI~a`2}DP!@L47%ah@921@)I3l~S1q^O(|^3v>mUqqje z<_UgUW;2Kgf(u-qsWce?JhREPRI-V1XmuBO7$tV*WRFoNCqn5on#i$1zLtvl8m_2D z(sgi-P@}PhS7}pbZxV*ty6}BC)yXclR)#E6ISji6zQ(>xL;?QX=Wi`|i88V6HFwUf zF=>G`UIkaP%G`h3{u~ckAB!$GKc#>i(c?y`ujz@FN{Up>uMGRKV$$(~Lvge^T{tT{ z+`CLp>G-7BkbR)0(hho-2tc14(nNOv68nQUQ06<5_)R4x4FBJ$yRE!pX};dfo$eG3 zoscOayE}q$L2~d#T@jK~d%KBq#nZ6|V;y6@95k=qT%KS{fGw^^IvsS{iU!Am5~$zL zW$rqB267jPbzMr>;Uz4I1@0S&#>A^yM`gS(vO%)h#}us+UFC;HG3$2DcpCXdZbCSC zVXv~mrf$>i)QE@+zMWhz?oSrA>ho|Gm9RU4dbsE7&km9hDCHiEbR?9M8?Z-m7$CE`x zRy&$L$Zj%R{lirkvkr$>8{;CKe&Yn#|L>Xn08*hjA*320b~}ux#J@p z#ihD|>lF2K)8g3uEnf2LtD1dQXmYM+m_V1jFu!AIx3!4Q9yK znh646VFoV@1+uW^pq@ZgkpXyXW-jnb*>nbaBT0D+LheVcW*T~a@KMMzmy3mIoAtS) zO^Br)@IIdzQPv~b1~E&|2#8cx$mhjUgs#jPs_F%Cow2^*-UDud!3z);sqF^1SduBX zPkG3VN848ePy|xVzcUjvE{RTl8(1Q(+585SB~;{$126N+S%Uz(RTagWoxnCE?++rh z|Ev=D1~t?J5EzvPGyt}5d|0z@4#7rqot66`>L7EUEb4yMgDVO#v~t{@@m8o%$O-}t z+wn#!O3DuV>Cjv!cUdKO6Kw0wzbol=NqoDFN)F-sX$^mMETBy8#KxPsDWD?nJM+yH zU4RrpYyB{tDXCdyfDiuYt1*^2bX{n1`E;07r>(&@?L^?z88>@@w~2G*Num>!OwAhC zI=9+5ZPG!n&412AZdiiOB{DAevg*%Ik<=B+MK*MJrp}Oj12b$;XRyhk1&1!8X+sP% zhAx7s9=sb7882V`%|U19>l+28ee$a#t5+&KB|#*sB{5Ju7t{@TL!tC}Y=a!x60Jjlq`awC0~p1l4WFlF0;pi%c@Qg73BlW3tG6PRk4Fxlu7RWkC6 z=t#?~<=IOLeFLqkIi7K*v;u{k_~4_SKfmC)tVc&(gC|!8tE9-2)R(6s8hYvS-i=C& zH;uhM`;9qOL!3lSmlu}yVr=|n^iQ*BV3F|*+ARYPUjS8qS!Xj7HBI&fQ6z42zezw?B|G*C@Tg9DC_54&exn?4st| tjOY9gr%MQJ=^#2Y&_`&&W}_TKhryb7k-zCNzcZPvJFbEJ<=@Rf006WECky}p literal 0 HcmV?d00001 diff --git a/2024/06/10/startup-data-health-starts-with-healthy-event-tracking/kafka-inspired-startup-data-pipelines_hu153a73840b0658eeafb17cde694e5f88_121924_720x0_resize_q75_h2_box_2.webp b/2024/06/10/startup-data-health-starts-with-healthy-event-tracking/kafka-inspired-startup-data-pipelines_hu153a73840b0658eeafb17cde694e5f88_121924_720x0_resize_q75_h2_box_2.webp new file mode 100644 index 0000000000000000000000000000000000000000..858a4c535b9bcabd19f500da696798e74a83ead6 GIT binary patch literal 48854 zcmV(nK=Qv*Nk&G_z5oDMMM6+kP&gpMz5oC)vI3m}D$oLY0Y0TdnMR`_p%nSXs51&g z^s*LIyRc?Cg9I~Y`DO8q>O8jk9)?8#Y>n-b12T;F>g7%L18@D`Hue7Z*9REBozP$1 zyX$(G|2yd$=|j~QoS!x?H(r_lTQB9lvOahIl>WQ9`VtcJU*o^&cu)D)+yB)68|@4H zhCBW5!C&gXwsY0~FaFQR6J4D4lO6xr$WcRV<_XbAJU#y0?ZYeQRfmb zBE11dFoS5F%90_v;c=DCyy8R6GVP&`y8140hnI9Z(TGCjeVbGuAqZ>vmeig1tg~k6VB!5~1W~yv-K=zAmVaj5)+#Wb ziXW0OTN|m=3dyU8f&ojRNEgbs3kh z3}QBm7x>eFvTKGu(K8n#__-8S}+QgNanNT2xc~1C$qng?Pnn0SE-0szm)Q564`(vJ+82gs9Y^j+a-$9h_@rHaf6ZRD) zl=h(}wCJxoT^neiP~b9x#8YZVxKa#OOi7M!?o|QzJ+lHFWKBkLyHy7S_D*V>H(o8l zI+8#an9O~&Z3g6s>!`NU+U{_8o@ghf(zpf+ZemiN?604YcqO-156jaRxP)l-b6OFw z4}Bed_YYx{y<9tuZS@lc3Ln5)Z*0>+b-2a*# zm@gB)W{UHVc?ImSLqX9f$o(6`TKq=Gft{>p>`N;)+d8<1L3 zZ`4HxVdh9hj^+L|6IVB2ZF8fDXft_g5u0>%9?&d{Z*p57_6$F(VdJ+9fo)Q=quC~DXLj!xlpIf% zpr7uRbHJiMO64vYgUC1(++;Hh_K&5Etm6ybyH?HudC+lmjXpG!EHjvK1HOl3 z>?_0mq|Kb|W|Ds1)f^b;JW&l-DHKqQ`QPZV&d{6b0|zGDADc(r??wPjr4g5@7PX`(v7L3V*(C^XTA6%ax0y1pG_@!S3%xI%|% zks-5+cM7$$l4f4%Uxw*4t+d%=f>#eFLQ^DP@KgMG$)U9e(UMXZ(QW-{;ED6a3>uDc zFyR%ItlU0(R0?-@P29#B`@FroZSi6h>pA48wO10!{k;R!WqQ%ZAsO_&!r?0DATz9rsu0uYn~ZStBLos z?$uE4KO>!M@=X7_m>@Knmh$&sgH|Rrtb=}+eeIE+34Nb-W>|54SiqL-NlxoZR2Gge z!>Vb5+-7l<-ftG-+TVaPtg!^=|HUAs&i#`Ll*k2nAF^pk#=AYFFLx+nNYM7EVdh>d zVXXnk%=ani!AJ&ueCTEE@~6L=hR7q4Q(XKYK|(>q5VflzMlRG^J~hIackeMpE*R%8 zy$sDrD`wTrZ7OdXlN0p1O&m}(T3wVD(aOeUWAUCo2#^CUB+q4iz7Evjfb0&VO^kf1 z^u0NdUaLN4-uM1op0V>)@`9!;$T^>F?!u?(1Z*z8Rcz=c^+R(qV4Zn{Wm|qDYB$bq zH;abGx7t>LR{LH-xM1#^*>T74RSz!ieKhrVct$}Z=;+`=UD->#$b z(L{dr==pPf-n+it$#?^t z+LqKkbC+ZyU^;hWP}RT7*^?JI!0nt!|DxcE+)(;q%W>!{#Yta)F6X%Q>urS@F}>&D z4bv1pb+*(rb?8SpG=+2ZQT7-U)oHtT6O8d`{DP#)(WJNQ*@YnH^rN>lX(Lo%+TJk02#Z|+r#uHV;g`;HZ(eF6ucBzW}o;7eL1^yAyXO767<4biVjUKH=% zM~#184Nh%jLoJM07JwP{22kg~EHBJ>0yM3cquSzVSn5-0M=o+8yN=JV64u?cjG1jI zO56Ba!#6hFo}%3=gd4i2Qh@PR1u(zY(i0-iM;G7}dIiL<;)X`%QXxu&= z{-ZR)BIEpvVwD7@XS}q_%WMc`1N^=0_?yWpk~&6PusG=we|J={YCv+7dL&{9f}mkh zkb7TQuC{>R*nHDZ)mIB(nx9qN!l)=j90@0GLs^gz2-GB`W^-N-P;Xd2ugp+qC1AKL zS|hqkd&*UZK1u*C$nYE<}L6V`(@ zap%qo3XcKzOlgY(O(OZsfHOa7lr*pBOwA`65wSK+eQGd>R579VWYp0D{qv4zT`Tyy zojk*SV5%|L0SA`?6qlLE%(3CfrQ@+}t3rfkL~v)pSTb~&(<$j%N|K*ZT4K<$x7;d; zR--MN)=3~P1p;Rg%UQjJC|aVZkZpN%%Fafl`l!DZo<$Qb-5fXP;nv!F zb#M4o6H1Ve_8#Yjom9D+t~jz*B);C*_~e!mBYPhGN|W>}(hX(uWuAaAp$qiTxq`eE zNKik9&Vn+rXpEPB*)v_Vgl+BnDPi`al0kh(72SLGd=MhecvmLsK-Y< zVC~?=+>G-1bevdrs9zViChenVlE4fd#{p8U3 zjb|x^vDk)9SYPeWewsAC23H}Oq=IlAXJ5|NS8FY||Jx?R0)uGUP-^52lL|wBKk=3# zYhJGNKp#KBB03;TPL>!7BgJr4PD0>0{KuwG6Ywad5EH_{eTSS)Jg*}Zkq1*$C_{*_ zwNvG#!(1 zU=&iVr@X6)lC2xS-0Nx8cPw+P?ka?@t|A(Ivb%#Qb`nuTQRPqmp)Tr#2HIrldTYY8 zzs1hN%$JPzLTA@q)#Rn}ga9|gc9PN>*(kOI^Yxz`q|Htf#F6@0$@fk>eV}|Y@iZXt z3rb79ili`f_Qg>lmi;{`JbPYAoQRu7e+UiQ-n@k%+POdBw$-8n#V?qgg@ZO_sxts z(9mJRFYMAX$>#0XlP=P9D(EGS*8zM>0T|4s+*6&mn@Q_9-SVM0e5$wV?Ppm6!nXPj z#KXd49PynMPcLujqF z>#Nl3d;dd<> zkvlb`?5lK>C<{Auv%){)z)B9j=Ey8nuRJHqf|2J&D=Yt-;ANMvEmks-z^6aCMwl@@E%OQrhoSE`Bpm>;4o5}o2nbmr}9jqWSRBg33{~Q6u#|pC&JIDH=z*G)}WpA@Cc=SE=0092?jI?&* zl;_0ScZx?=@pc#@Lo`m3LS~kD?aBwvcb2AdQ`fPbNQeT$rn(5~lo;5wL0o<#WVpzr zgiOws8@Z{JL%!>V zsSsHq0$FjbZO}y}$gb40&?%v=onx$W^Vo??vH)So)aXmI2NUz?B4hJp1$Edx%ucbU z2b|8DGDG#X3gh7*Q)r?L+0vvDnJC0t!+|DR`htVtB8TDho+nGO+^3?*8=Xc;=M9}w zpZZA(#t!SD&Xu?i3W?uhe6blIq5FgAvjuCEMUV(~yd&%*w=7!pR4b78r_L7?z?ChLpngq4S>(H` z@$iqGI*_~M_KFVG-v|0MQR*)aPrv+M#|q)$_vB2<<*z`RH40K}l;@J1V<>}#zPL-$ zi78sj@|WB}3X+O8a{xWh4@EVq<5{dG~U%>~nANuygnSkgBgP26wCxKR~*+ zxVMyeas+3LjQ$VQ*{JHYk639DG1cSs_huj|TZajArGO7CJvAfPS{?(kJ6+B;7N78V zv%voMRbv12E~6}364?twNv2expmJuxPb?rUpFj-nk(@$GRRgqSlhZL>!t(tT&)iku zQcLCc-uEWcU{Z`wi#`E(0z*?vT~7Y78sEyD0@rb0Dr?Ypon{V!i#8r7i1b>m(8pPh zQqcypT%}AHiIKkMOt0sAttx*EzOYoI_s+cN>?)S8CkP)O2Uiz)`ArJY1K^Dp$>$-M z6wOa;2!uF9i2wai@(cJ6xaxEeEsYj&)fnCxuOO{gJqB!EZqN|1;o0+;8OYo(cK=n@ z(>^Wf&;9tWuXy&<*GY@h_(jWOzuEfuBM8*P=V#K9XCEZuuD>yFKUL3HepdcnGYrJ% zwrd)ylo2$)^ulol!>8=}5ro+crqjcit0%-8pV@4Sl0G)hL^uAIMa~+^$B=bcwVn5XJN4`dAOe-Y3MP75L})mY2&Hqos@_O$>Yn zLx6|~U*N$Eh2u!t{FlR@q`brv%Qdgoi6BdA`p52w9wwq-*wpb`6Z{3=d_(V#|I@GV z3>Hq&M2kQ#a<POw6^YD1B)F3;7JIY|R&2mtLWE_NHZP zGCc(xhRpW8?MD(6QZ+#8^wJ5!rZAv&9jFns!9RZ<1aJ_hvGG?X%k{A9UK(uX3Al8M z{@oyMVJ}TirP0GGsE@JX$Ri|XOoR!EgJ_gj3%sT)IG0u{$JpXhi)a^n_P`%yG6@kq zz(I{llX3Y!Da#{yBo*dsgdgUcZpxV&7jIh+NKBF8HFkcZBaZM(9`~e_X(mBCj&E6k zcRQaXwc-G4nPQ?>-rN`u9K)U-e>IMjwGI&RqWMrcM-_VEu%OtnCJI*N0P{p1a?mb$g4gS2p7v}Gu* zF2N9H?>qN>2c)Flgo2g2eSr5ma}ED_)thPtT2yc@e!GAqXH+ELN;SnDuo+R+`m?!6 zq-=T^qz;$V+Md6P$+Ua>VuN>jzX2IEurg~9+|f>Nz$PRbwXeWbcbgd6-=HnREMa60 zwYarUc?oz`b`ZE`L)VAujNp9cy4!7@0$&M%gDFDD<+`PFY$c? zbb885V41f(ctuQ(b^8o6i8st;7qHs`BxjuPc4|||W09Jqwi+6dbIt zZ+Ad`^BWLgv2NcuX$fu+4|AL4QQ|?a(Xx#k_cu+06goV&*S?sr8qLN$PUHf~lpLAv z1!m;#LqlDJ6wEFTvu7RG+@CeaUg`DVhe3}o)CDb5u^G|P>A!_Wn5TB&{2)J%Wy4Eb zKBZbr$HA{ifjqKtZ*?m$8D-3~$9mqkCI0}!nku17N~tJ4;;+6LI0R{+98&bR@r{ng z$z6IfDT*wri<=~xM>I}UiDU%41I!nzd<|quvCcp2gR_tB5iGTO*@4=;q+wtbT!X{; zFjO!APGMH|OCl9ov0XRm43jGF0ScNdjsKu`q+Rul!%!Sv3x92ZPJR(9U9n#mXVRR7-yF8uU{0jyxIG7XWk^g9K2(k9qc zJIDx0OIPWaMA_acVkd%01dmp5V2iG+ksKKnSU~#Ga-^>3Fio|q0%9JM6sxUuYykt9 z1F`4fkU+bhLx+GYQVT^W!({P7@MCR^`dr1SkAF?5LZ^rf^J|1JkfOHRs;Jl`8F;TD zx)DDiLDug8R$j-;70Nq=tC!j#YpS2TeDP<_yCw4WD>A@x?&-YSf7 zs__H(`U!cKjETLN7|}%}^iM3$?M4gM@AR8?J!@fR4bKaA%4+5=QlLG53P@Mce16Qp z`~-t3+)9y2s-auf1Ir3?*+^9>K7c8pNm6(wr5R1{YQXt_ngr@`U8%LtZ%*R59oVTg zl>XKM43gFQzh8nktlu3>OJ;zYlkD-$ZF(&n!SeIrX11fXcK{8m@xJ7>CXPweARdVA z*az4H&JPz_JGlsl$AB|CVQ|QP4i2YaOdiA$3BUmV;pYF5>JzyRrle*I%$DJF)*~;9 z;abcvT6L>!JOSaZoWZWA@hZiIfH_ZO+jdD#ZXrV~(V4eLaCX+eW#!}vza8ew+ma-K z$uoou5bFQ-u2QZ@*ykr4-a$)s+Gf(bgzbPhiW)BqBo+E!XNbT2Mu_>NpC7E)W})Ki zHe2Zr?9Rmh*Z&_Nj??bsJ%QDtR;JKAlGIq>vdgs7 z!al%^(+G4lL-xga)rqE5v|+#+sUI!pCHhP08wqgst)#$3FyEEjXV`A33*5K{2}iUCT1=b()%H%3OH7d;Mts!Sw_ad%uPu&|XEy5tMrwR*CLBFCu~m&N;VZvBe5aQTNl zqDX*jdPZJa>+`LMie4cnC@n~phnc@Occ%xqi^Vsc~M^6Hl*hWnT9JU3xkF~ z{!0+9H1#X%DWX%%inHO>Z=Zt-dUf3c>Fgf`h*OYRowSzRY4D(3@BVaZ*?% za^T`<;mb0PTf)&9{zcb|KN74^ zu2zLi48j9AHBt1hX2{wwX#3CeX-MR3*F)~WT9vvde?JBc^FQzTZ#kqE74-A4G(Ku_w z=rZ=zCn5-$JFM;j@G#RYOq5!zTV)&Jf!<)qfcz?kmPTZkVOzzg02jp40Pz|FsDVUY zFDp9%Z<4dzn-8X`lese2KnHoytxpZCRhbd5?sk`$gA_eb008nBP&F{q*Ujj2T}CWS z=Y9FdY6_6)NZiIm@nDA1gH8k*86eVx&+o2oRan4Yvu$qshfy1+1~;?u$I2`()Z0p(eW4Sf zNG4GAI}wDK`}jXmV(S}^HLm%I$3Yx0e1|Ys|GfGg!cL}V)|8N(-sk3nji`N zmZ{GL=8X~QH!G2HB!}IBn~SQ>cRsJB+z{T}gSBG9NM?WwG`Gro+`uO1&SyLTPr)Pb z>Ef`7Gao}!r@BVKDs^GFz~S?RN4nFPx@}gz);Sz0mD0^fvFYnec<^daUxgVki2P7* znE2aB=e7jCBEES^f~`zrk0M~Q;@H>t1*VYXC2!==bl=L~N8vVCu<>%Dc;S+R0)@3@ z`~h_dl0Z?FY|H4OQTBn09qq=HLae7;3cX242rLD=zd~8&V7b}igUbxfVci6-(E@d zHz`|fr~M#&Mek!CpXdU68e%X>I{6g}s2T%fYhxE`MkLaa{8XlE%#QsUusA6MY;5*7 zL5)@V$x<Uy(|@sYx~qQ^kHDNScQS1`)Fr@?1LdIIbl0)v)1y?O@tE#-& z%c6I9#6r!1#&+bSs&4o4hN-YXEz##E?R`{BY@Q}-V?ebM;djV0^#E2<3__%zZ3Z%~&ExyW9&ITHYaH%A zSRFNuu)}D%@xM~Ynf?B)6rRG8X=h`u&Em3+0V$B2Sa}%XjH=am9BA69{ZmTb`|KuQ z(iPBhDkX6UzDu#RNwjxdE>|!p@gEHU|DGo#(;yS8YI~1^!lBRsYqhZ!R0!JT%BjnF zI@Vvj1-G6F-S(=p#u=-Q2uEG`gWb5E1v}El!us$rdzC|&t=wm|Xl3&j-uWJStO-j0 zqag>f{P?cBTACRz5v&CXU6_;>$@z-O8j-wvgz6)AZYAM5k8Bm%{{?|ws?`SYA?DEJ zckw$lMY+N)Gb-nY*~gG+qL)2V_6ww6^J$&By8qV@(bj695$zeBmGbs9uy{D=_&AK?_t^P%{U1F_sji*D#H zGweplg9@--CHM9u-g@_p=KRM0mh~H5edrrLxl#V>DbIM>x^h9DGnOfjs@usoLF+1A z{y*$nW*-8#zxpUqsCz-;X8${_D{;4i@bYuYsBmNBy4%ME@D`^UFS#SURbc=C=CT$06mv6tch2L%0cyQ*+@oseF|lt zkCImOSF4nGrr}{Y3y~^X^Z_{?Q?iKoG*0Bb_y}El{4mej+i>S}rrK9Z`$SfHIYMde z7mLsqZ9UTZ_UD}tWRo;QlOWyvNalQmb~@8 z(9$Bt%-63OP@W;|Mv10%i>SXwea^n=eKdLUXX@LyL5W%{p<4{jVq~P zm70>I{ZC}8gT;LEpPHmw^DT3rO@I3zrX>=T*rksB&Hd z(|RFuTO&6&rypz8{j~aMW=ec5ghyP(vP|1Lh}?& zwwEHJQ1}{aR5wq#^V>V2dqf&LRRWKoDJv%mPwbAVJS@jcy#A<=zbG&aJD(r=WBn?f z#XjwrV8gL(JqxAGi#-O`o%6syCfre_!6&H{XeRQI$}FjeqRPndfEsy#9EM|(t6`yy z_P+d!GJ9SlvRP~VznD!Z@OJW`f7{aJJV@tFfH=6z)6zP@I^`NS2%M0(UpQ>SXsO4eL#@2(Q`jnD zuvg<5Ce`cvp`&#pXd^t`2liC@Ygj!jT?erVozW21^j#ZPb+`U^y}#iQ(q3H2eJ+f% z@e$w1rA>g;FfyxTqtz!E+&?yLcNnyP9~8G#N`e%Y4t-1&iXsmqcPyN5Q?7ldkb!D3 zmakyO;$jG)0871KPRf>Q7i@Sc7*faCP+Hc};k>#YPbGbp5OD00QATrK=noL3MoL4- zl!`WO>&+`q@%q<`bUc|PIl5;${}*Bb<=6%n3FAt`)DJimH6-49Ca+g_R-^fhCo`FU z>aeC^j=K=7G;-7ai1S;TrZ1nXIzxh-1JjW?qpu z%S{dfzjO~37G$r9^{}5qyr1W{silkP%hxHEDxC@bcnK?Sy}Z(w+0G)o*#N5irj^cw zeFe)>0TFEen};H?AW>-2m8V*XAq29->EdB`QZUoNIL2($_W?S?AiiRkU4K764P2Y) zcY)G~H=f&xY*#z8JDL7RJcNz|9&GQVlexm4^p zEre3d5oO+KTZCa>gtr{o#6E7WG8r^wUAXT3jlLky$|F|1rhgrT_qhgrx3KRzI0&_1 zeMIP@y}UcZqcN#xMBiHf>^JvUI3bX0i6M?8zo3v9kx(|P*eQFgqs20IWpPngUSQ;|8@ zy_=ApsFLiAgWe);nVoQ;Dd7YfoO9zK5|IG!uq{p!bAwCQrLzH2$J94V^p*0#>ng~I z^UBuP@@QJ8>{lC09q4{4Y9r?C>;q|Y7jt?q+Cfc82Xw)NuKRuDMReiQ5#Je;8Fi7n zo?S$`SHFB-bg3sj3HXX`1c=aYFEy1KdYzY%ImTO2KC`HOS$qmT-pPv<71xjKp8hv2 zuekYZzcBXk8f3bygzC(Dw@Xf~`bwe8c2jG6VI=pcdY;SK+MdU;E9SYQ2HMI_sT?kT zvLak`Lpj*RN_5)@u-1$}!n?Jgi6>k`YL;W3lVE>kj0X69a?Swc7NQPGq>%V{ z#4Xig!^w=zZh9T6lF`}L4eLcFbLPsGIu)1nvzt)jTlB{(cgm}uYa8hDJ53Z{^8Sz) zqq~nn8)Kj~E5)gafPRs|@3T!|nke9eJfQs8{4?ri<{;7B=2bWotxjYr=65~ z1=xlhil);qtjC2#Im`gjavrw@=BC2(rI0m#CeSuxxQ?*g`pxMCv|tHRaIn4{8r=k~ z7f)-f@n>hR^6zBDU%YN(s^j-~p%!l`l|AC}d>Q>%4d^XvMN(nsOC=wvO>Z-b7<!)r zz*ltwz~yv0^`m_yTra#pJId@b12?*hfwa?2-@kGW6ayglab-VrNft;&dyZ7kL0fJQ zrM^r3wZBu{FlGJxLQZ~iFJ%=QN0M9X;G|gW*s7GvVIlJ%qo)XC+7Y$8| zk$|D~(5+NBeV^&@rW-=hDv*G)-jD|mesPaDQxkYhi@nW~xWY!KE9W(YRt8x!G?=6R z@x4-?^fE*(LV6Aha3P;W#}eYg-(a)8E7Va?O?cwF4SK;xl9Y<0KjxN_$7Ov~?gjSwT?xzx)3y%=BocCTRi?-M~=#j>DTMQ@tM48uX= z+E^2O^e^T=m(XMVZNKc8yiiaBD}wR|-Eu4??Ma{qvui*o6$-l(!uAQE*GwKv#Vmp5 zz0xFr@u=B>E>-CZMMn+eRZx>vVY`+-#?Hk#zyhSZkj(CyVu;RM47o)}NiotU6p(s= zi6Sz$1L1W=Z@?Z)uw#}$67+W(%&qYHt2Je@NdKs1``!3i{!SR(AD_u7EOW>jbMAz< zgR>I?E^UM=1V>A?-|B82KO%|J!Y_h&Pl)gDp0oOR`q&N8IwDXpUW3B^=?ct9KS~qXBg!8KG;p zK`xj#E^E2fA_hS!+6y4J553Yn*q)=RG{Udz~3i;F{;V7GbxS(AP`xkLoSRj1I)m0Mo0 zpt9OnQ5}$%)j6?m{ouuJRZKcFB8S=P1_b5!{6~m9iE}c6HRt!4)7!+VJ_vGoJXwQe z`Uyk^r+;R{$gl#|FMbj=fi|OU&x&b0ow8ofgNux@lbS3=>|Z=UYc2dY7+ za}EChC7XFgC{lma(xv>U-`^hmDeLw>K8tnl#JeVDt#W14I7z?Y{0gs`*&ed;9Q7PW z7$!B>WWuD5{6jYQd%eGL&s#2{Ga;i;z&gx}rx*1)f%`d$TAF*u4a_$nK^%nyeS{|} zr6@`LXBR;ljMOLAu8)u(Uw((P*E;U6y!R|9YHa!zT!{Uz^Ab8G_3ONb3sl)AIGh)_?UM$tQ-Nj}b4gmEe;u8@8{|J{q-8Mz8xopWL5XD3sOf zF#gTu_;U8e)yRNE8|R~$kJ zdc-xjz7lch9?id$H>sB)!M5TU*8V5p;G&5H1OQ~H#s7RTlzL#vdUQb@{(I8csulm+ z)yG%>$M>c_gDM#q9XP9Ox!=TWT=WleF`k>`@{^gwB^K^JFWJt0woi^USD$9rG;13# zPDxS^m$t;S)KQQ0f-ZO?3fOGA>*;g`42QqL|2AOQi=R*QXQKp$ zo*!$do~E?39}i{L4z&nwD>o{tT3(d|XTNRskG+LbJQFQScT~HWE$|Lb=g90xedcYU zlrnkGYP&uD3V98Ne7j=-T*D~x-tbaZ@kv(lULFsBWp|`)`)DrrQ}zFiCLU!%_a?@q zgk;!pYfJzli413K_6t@JRI#zh>Syi^nB7Y|FzTQ%QNoZGi78D`odnm;tQ8xzn$1e- zA(Pn#Vl3%073>`r`x$=7ouLQH4@)gBr&RD-fdB1EBKqf;?TCaFvWDSAOCS6F=B$KP z8ykhWt{z})4P&i7;+=Ce!#$z>SYwGp`QEDz_0?GCS!!OEvmg&$@Eab5psQ* z+*Nvfj=33iv4dCAz;fWyaq+?d!Y{v#7jUVCGr&(1m}g72+2f3dnC?x^$joa8 z4NSx{(ViobA|_?s9(p_eWfx1&qc;AY5qi5&Jg%l~lTTVO8q{#6H@gk4QFcpAUB8JT z3?@@>t%ZpLkY_SoEc{voJ$@o|Jf}h0l7`){7xJ^`8)9?Q^S7ts;~uH zI?1F41Bi@_hO+VWYwN_)u`k(AksC}FMh^!B@Z?d|st^hZ=`PF}UmeFFb!L(T2 zw;*8r+@9vxuRtUY^)=xv7PVEMljWR=*iI=gEr4iJTrsdn2RI@lqAQ|p%qWGE^}WrZ zw8&3zcbI!I%M^+r?PG`#5i+2M{{-m$$I@a6HYmSQIJ@cJwnigIo&@6speEs@Nq}q> zF%f(5q~;X8m?l**;?0=_4yP2f@~b4g-j#Ki4;Bzy1YoE$A$D3IwXEA-sv&sYgBvJT zs@RObtm5y124fKuzm`K9`7rBwV>!C(BwtK4X;ZA}jWv;Eng7VM3*X37^FGv2glMwp zm-v{#fsnt7j1yjQ=We=46gUc+K0dY(D3me2yxt~m2|)H96dfn`nA@o+Z*wH$u!5Hf zMzS%tx4_5^N>fW?w7|#afi5x)Qk#(iF{apNrTHq7-!Hl;i3JInmbZr6kw;rKTF?=Q6}-{31bn!>S_TW@^*&?R&vDH2=Nm`X7zif3jG zxvpe;#4{pm?LA2_K^M&Z)J&#_R2GaWacDsep^yeK2R#U2D8keg6h+iR)T)v%yjOY!}U;ucwdar zAy1}!ZX98b`N%xhnXji^i%yRAF${%WO$Z6#8v(Nm)vvRuj%^J?a z!qDjHk6576ou=*%b7yA{8|2|ngq}ua4Ex$3^ylzoD^K12)ym#c1Owxldw{)=1N&gi zV@Q}>sLa;$z3NG{ftDbz= z1cuXNQi}K{wWQj|Yxkp;XXCmjwZ#tJS4 z+$M2(=6BjM4>*^j;du+C*dJm&0GH_P;Nbyjs7C^9c)W?v&j%aTwg+d!@p+ME!zU1D z&xM5jt7MqeVmxkRg4++?1-}Sr0}JU-bG^`g;g*Ae9r-_1|;=A6D7p=s>k?X-E-H=ccIN<(VoKB2gV=0 z!lN~X9e~D%A@3WF;ZS)%B>|zw{wIC4#}K7lp&iHWuFd7>zr_=Lf`gmhefAf@!S62~)nqygx zI3FPp^_Rz`g~1{{t?YJ#DdkM(jg0bPoQ6 z*lo6|8Tj(U*pZ6i(?s&X1GBQPas4EOv3iN*a6(Osu(5MY3n3vC~f7#9{%=$WQ{>D{R#-_0p8+7tASJZcXv52cEnyo<=KLonT?B0xodPCbQ zINt-+&Sh$fGY+hit&{v{oVhl2AxZNtq**nH%*?dij&o9>Jk2LfPR~wk+8tAqfaEY! z7vK7K%t+JTWj31vIT0s@>AR(^x^_mb1>z&W?Z~X_4D!vXdSx<3OlKj%fF6-!JgNg8 zyhGK5To$^^yeHdhn@b{@M09J)QN&@qaet!GdDCSyy&)Bn}*unJ82i9c65IEAf1r9{ z+IUqrgpt(uPmR5bxhDfBIZsVTwx{rh^S8l;lF>coOi+gCN(eaL|JmMmZf6-!+?b}! z)GF`gLTicXd=u(e;^Ux`O0@#3vUF_9i*Fhu2?2~kMr`?(G8v`v*(t>}&gVh+DqNcI z*`Qd9sMX_T$W=8o`YdUjGR*v>DPGLnl)-cplsv_BGj2Z*l)h;@S3ui{A$#M1=uh^qF%~fh6Hb;Cw4DP=yMxs&xidFvOR+ZLlj4 zKZH%CLZt#i#I-JrII@i^BU&k7;QeJEHTYA(Bi&4=HXeoK#VYHN703pqx= zf$i;DOqAGm4wbIo6R#XB>KHb2kh7Fpl|(~E{6VauMRLtLjjQY0z;2cqO?dg0XCJmj zu+NmvJSU-5wbeLS&U;QjJ3?4zf=ca&s%wWwguS%-v|)mFp+H^#U&ANq<&tL!s9g4P z>(uMw{*|yZdE7z;0mpLZ?3Eh+lkH(D7dZH568@n=w^pgi6QnWkoDaG+ zblE3BC zYkc+&YhK3oeK&A^-s}+~?+0y_>NBEH8~NlE`E!|Xt$^h!teM~RFa^&aBtP1rxgdgX z6If=rdybH2stP*4>!9YU#={Q;YTDLrxnK?I@Yy*@m3N>cv78z4Sp@fNPc7VpQ}Cof zEb>wYlU0SV0ecW7=8&-H9}qq$#~^{C2T^;t>1l#Kqjbh~A#G|HXzx((lkMuJ)}9dM zx@G!gn_l{7=HjbYI`<=3dyQNjkL9T76+-R&sD~iT?mCmG4&Ju0Y*6@%$TvBc)-&J2 z8iFrMxz@1zNIV2)mAtN+^pGtE9sEx8Gl9hf0EL&NgLX7Q2uyJR^1H!5c6lw01p^xf zHsQhhc0G+LAC(klq^~Oi!$_YN3%?VK;u}@)A3ELWx{~}(u;Xd^!3`jkW8gD6m^?K# zw{xyw-cd^RnCkbqMv@~Y?8QFzdywY77A9@XHs;L=z4sb#)kONIU%6{61u+Fyp*(HI z+sZEMd)@^Li@t^kC5uC>X&8ZFsW7OiN~t-CE*&TNPcgn9vn6-qS6b_jTHA<30;v8FeUD`y)TBT`*GBDHY#>}y zxhpciLwDM$4cZ`vsGKn{w6W^A>^&Aajo0Sl{*r2bphR#rvLjS2c_#EO35aw& zLBCdqOfrRzL@yTu8Nv#%(y#G&bSdMCC59i?|8A;~$LI zH1pu)q@MsUhSUt5ioSFn8J%3xW9LsN=Ji`iZXLrzY7fb_M|SuetZq>znf@iee_+({ z7FBA0DLB&5Sq=-pxQmnF;cA$uB_S1QCh)oEK+%GglUqBUciHmB6Cbu5 zIV>6VdV*8QTMZUHE7waQ8d~zsGLGABW}M49G_vkH!N{MAj_ekSj+%im<@!#WG*~ko=NKH$WHyvg6|lvowLq&=PHXl zNv1LBV_||Ea1SmWT;P-84|v`T^omDXWyTwD_#mvF_)6}^Q)VzWNi0-Ixo^m315u~% zBQrFr{3!i&ASrYmzEpa^=`kTq#Tl@dmVx5UU_1+831ou~hJeUV=2Hey#<7w1USB3E z5O5G}L#E$czwdX*;!7H!NDFTgPAKXxC76odAgUUcTX)LN1!p z$fZ@sN(x6q-XCXe}OVVT*U{o zSZmnN2sp*q7k{*+TgxaCLKx}M5kOXuA%UZMBIRxt74>=<<@)8i#mP%W9;83?WiHF# zQ;M?}|ICk0#(7mUG`*JE&#i-cm*%59M!DA-a}hLndF%*7v@IJQvOOXHT{-=bTJ&!m z&+-@}`!%D6mt7Dv{tKaX*H>)t@IdUJz;&Ee8Rkp18DJ~1wdAWeK&I?+psz1>&RklI zT?h*vciIA^OnP2sliD>8TjVBImpOxVRRwjuCJ=vv5qLpMKpcc0ITmdQk&h_l>|6;& zBzi0G1&=C=!<F%a}~iv_bs5MDM>Nrf{q9ZB=05U$FU0RXxK>;)#_qz|PEi<5n; zm;B``f1#OFpFypwskT)4j|ulV%oV^5mc{HQe%tZ%^-?>YmXnbAp*lrUcq^^+TxPBq zjq3Xc9$n9xq|!^?9BMc&OD*9KpYDaEi3}?p_+zhAlAUDAPh->kbIUQkT^qtAqH$tL zRX~ZOZ>-fS{3{rFy?1VG?FKb|kRH;64ZmSv%XsL~g%@@eGkZ%dM;JvV)mZDw@5UOW z_(`~1<0Q)P@emGP(`(XhfqNPwNL&V?nEeg$t(xiZj45+(KOg&d<9z_f8I4hjn;cT8 ztBiJF5*F+#ZeoQt7dbN!T)=EV6Ku2}@@D2p#NnaPqbQA-9k)ifOCCd;Ji_N?=Ex>8 zDPha4ZFHyBx5wx=)(E`u3jHiB58*e&^dvWwK9x=BB|!LZ)GGa3hH3x!Q(XhVxxqL) z9L1S^tpBqlGYwW<7@$o{OP+^^Nz>lBP+1UF$#5LrCjJv!)92ZO-v&wCrb|5EB0HV! zvrrjBSzf5EgPq^%U;-<3=E_8Ez7&h06jUnQ7U`>Yb5e2*9?uorTQ?BSPTg@T>#?@554=TXwsh zqZq~*WH=h*yTkZ&HC+O%Ob}$x9PP7*F#=y$$F;t};2WsSz;aYP$nyW~AKYFmyS^Ka zhP+A;Down z9^Q`}Qz+rw=>$DaRzz2wHl%?(F*LR+^jbmIuQM*PjL(y<=~%%7kpZ2(_Eb}7 zIK|h!v6w}!a)$UN+J`s)QgFUg!9=1I&WOYW8*uq9&~8#!B&5s|?RILDpaR=f@%CyY zBfq`%b~o|g4YI;{iz7f}g`(19wo>QtkpZPpg|QWq$PGax0E_GnyVut3jEVDme75jB z`qc*+&;Nb6Kg(8^rCOJ7>mr)U-#RgS#M02A2<=Kx;>ZhSw>oEFW9HTZv@1!*7nB-#ZH6pD{F3Lkl$VHp8BvrNQ+H@2DA?!UcJJj%UF%b|;s z@XgyAzOvS>Gr>dX(uP)AS@&oIX;lJvgVUoni`J5+ymykQ;-0<2*%r z6P+8lrn6#?Bi8gEwLb@=xga!Xpon+2+PG;w@ zlTjHoWL|ndgP3~D4Fd5LGU3yGO4315Mm+TjzxDpPwETui_bNl|sA3v;ZYS%qJ&|C4 z0W~c%lMgmp>djH}Jy#B~-e+OD9PyD7i7g>->ZBx>CWcIF)qUi$PssI3QZ9zh;-zB zs~>92;@2u9Q3xy}O3}#=@i7S-YvlQ2z_aLJ5Pv+^TLBtre}HcWWDkm5mL9`8WauB z#fByElJs@1OPY?lV`(`Q214D|AjNHj4%PIB&L!XxWsP(6ek0}~kC0#$ggzyF?$emF z1dNo0%BA`8@LLrX^&4IS|3M%`gYNA*KueA~M&9z@z+t5U-h!^DG0bwdBG=X4Tl(1) zk-GOhE|P!-E+`tmP*u1%@BcMxnVx*rw8rxmmP`Vzh}| z%Cr4A##@Pk*(N{hM7fO-Ks~b{xAnFENX}cu4PDaS<};Kcq0lJqT%L$rL;pas>&Yey zjD}myn-63{jg+7+0hgr?xKkR5)@tYWS<}c%^nWUyzsmLM!1l;_WIv$RMG&P(rU>;^ z=B*7~#vQ*N=F%dcPxqlEG~WL*eveM*bcJYWGkjC=f5*N)rwWN$12{@_7@zWNAM$~q zekvisy%v{bRR(4nmW6`$3AEK^Kd6T<;rh_vxsP&66+hq8oRVR}$ezf>}$^1Qme*LMWCIOZrU^nfyW7$yw|eF=zy|Yi-E=ICDLp z7)1#a|h7BI`dbehoLtyA3>xqw&X)Je-&u%A;6%}D zC(UDNN&jv6;9h|ch+Tj6@spXBTjGMecc9fn(W=ba_y9yq5J* zcpb%7sX(+WpE$OuZ#e-m4pY%Ob@nCUKpG2tnQ}6R$BtqaFU!=DjVxX3ZN93QSDbE;5kxVv zi@1g~Vvf>lmdw!N#0IBw+n(6GeXqH=c4TVC{0py9y!J!7$r@vbwYque2d$ntjwJMy z@g0F-YnB>sjj=#rPaUKla%Cr#FeENnByR`SpKxm@qvh-$=95u(w3Hig@MMkf3gv-euOD$Aj;D%O9MFgFQa);lWtn9_5z^rms&-sZ0#{I*r4!!M{Y zNsRb_?(BSN<#%m7BCS=EC;OWp9wXxG=_tn&D?taAm<3nFGPMOyS?F)!UDOu%7 z-Ke799oC7!Q2@lH3oj>q7dz1?v=7&YIa+AXVV}HXunWxHrM;r+l_W_*1Vm`$ES{v2 zi%@ne-{{U)GhF`$`+bE|;t_F8p_?moA;S;l*Pbm#%OD5yxV;fA;v*8%^nydE!0$-$ z6GdiUU+uJzHNPfqV|X_!Klb>LNqZ%Cun7+P$r4676~@W3#IHmx?)jPRx32>xA(OM^2jvbzAB3t zEW z1zgQFd%gWS^M@SXwI+xL_6_}Iw_AnaO}P|`1mehq-0JB=l+ImAc4Y7RzY**%gbv}$ ziB)&(DB9zMea!uGN+&YYz+wQAO9JEVb@M|odNM?2n5&J|wZR);a!`++5pDd`54GLn z7J^aj3IqG^PIvD{JyjTZc{;#AL5lB8^*(H=aWaxg>C;yYJ(l}xmm+Xv*KN!PH+ZIx zC$Xoxf*Thkl_9Na$i&%lC$!1Za&<66sxr2nG;`C{pH+o(++h-u0K)GrIzt(kPmlKiie@2zG*bY2M`04Pegb!g3 z^R>bQ&~C48xh@VZnPf&A=@BO;^O|zLtZkl8P=|>sQVAy`ybvnBH%2EyP}gj(_2geF z%$I6u#KM*gK?$oZz{WcoWj|1uOY&6l>#>LCTJRl*zj;M8T2%Whczk}%5RBe6 zujL5eX6_-g!1ro2;vH42!p`O@l#R1T5bnh!>Rzy`K^ql$CKKnQP!q5yOiecdVNK3h z8FmDrn+NCoPRBASNU1lYewXo|#zcNEG91P@82s|p);EjcF>kUk`ipPjFo{x)8n9ZlQl4!@|0jAIai&fJnc{Zf- zl|h6sp}qSOIKRuPwBoLKJ%A?DzA&|j@;`;6Isf7C7^}?zv{Yyf2RcSv6;uq!jueym4|7BP7ZS1T7O<6oXB0M6Q$H~-%gb8u!K@tP2Yr@l&T@T ztP2)GdGx$f!(y#NNRjqF3+6g5!pgRoY?`-seUiTWzcORigvj zN6Gv=)Bvw86MSFAe8B#fPzX*y;Q^Fml{iq@qriO_aS*y^xhHeZ1|OLjUOxGu(o{$z zX^#0F4$iH4Xp<4#NP?V_*j@46nB~Wy89ua{;DL}?n7O`0ZL9tMhkpi+xGf89g-Z1* z!xBh8dO}G+j$T1@o?i*8K(m>NaG6)}V0&WMuIkZ6j~M~+k7TB8t4Y~vE?ug1m_kR* zMm_WG-9^z}-9!W|28^(WP=0|jhkN+qJ=G==xxZ$MtyMkwubIf*Y?dm!YffbgvX-DC_4F6u9L~YKE*y3suvU67n8ivx z!L{5XuPaRJU@}IQciAN)!MY8OI_C@K-h~a`x!wya-L$MPLw6kxJW(&%FG|8?S9UTM zIyn`vMq1vlEXvTDB2*A^5zm@hs7+8*=G^Tc3CAed$oh1<%#{|vTO08!{FefSN?fWV zA?D)ZPAk?h+uJ|$SH+)%Nc#}FZi4h&d)XYOf8n&pkGLCa3#0%A0P|eRld$hheI33p z+h=3mf>c)`_2Jsms(~jspUQHxwejTHvnu~43$vAFaYlXe?LS2bk-Qb>y;$iP4yYxr^vEU)G6#1`O-ALv zj{9MVa&Cp<#spLy4=~SR_-&`B6SEorgo@|dyyKQ^Xh8%^76i3E7~oKy;TlkMPvE-Q z)B#7{K**t2B_$Bt*4%XAgTTN4dJ>2%OZlGc9hB-HCtbeH*SC(&MaD%-v_&F9%pECH1jgaGyRB<=ga=pnsN5%&PtICu-FOGiF6yP+?9o zWF3@jkhKh<>KOt@B1_$}6q>W}U8MUz$?<^T(rV$gaS~PC&3oz+?TO1wQRH4lZC_~m z!Oxn|N6AhigM^PRvcICh_DTe7aLxNR-|)YGTBN3N=dkjf4txVV`soHp7n{|jlt0DP zD*r*v;@7W#)o~l1xO(Du>r1&wVk?`izSg5g%eg4Q>jm7P#^EA>^t+3o0Tgc=wF#`< zI)?_7ZSJ!e2Y=SQX-F;8Z zDZXk2$M9kyLc`{tys$ox-9PR_Ix6=~>8#ol>f+K)33z>iuUxOLO<0B7S9JRyuR)u_ zr1pX~tG_RNfWjllGpdN{aa%zeSAW_K0tv6HVQ9?>9g)s=V?R}aOpn8)s=aBK6SEo zePzrM!kQ?HM(#_?c*#d5~tyW^j^m zKEB8G$t-QV5J|E8PIN0G`d}>k(A4wCIg?hrzcYNHeLzT&Mm2|($x1rTv~#n0CH4w#KIhxv8@7us>Ofz zqc-OEwSrv*07GOGROCXKWtzO;tDl6WsT$LBkW<~gYsz>|LdH1G5flXeElPJsg^lr-tG=rEKI6KJ<#vV7aKhRt3Y6(%{*&4`jZd?C{`N z$t%>2FEyVpATLq3*+q1+3XO1dE{q7+7Oqi{Q(#g3y(2`{czRCv-T~Mz9<*FQ4)(0**H)WXLaEBVV#TZQ~$CF;iB+ zo6op9-406dZ2xmHh&)u5;sfdoA$0IJ1~ejwFg|d5!Fa`u5XbF!Kn|wc@>bOFynGnc z)L}+bn)cJohfva7@@mRucI+Cc?M?J)Z|gc|DF@CL;T#?}Vj&0hS?xD~9g9<Q4dfeaPm8!^VYWdET#&K_bNFjZfx^KWv4ZTIVbd%l0%%T7A*$$D zc@C9wD6z8zCq<`M`-+QGdGEsd1i?fqCLdvX)O8RJrJlc=N(dAGlI2lA)x)M#ICU5g zHdhPIIXtzGG%cv$w1MfU9l!bQt`2NSdYZ%rqRR9J_Zn!d$u@Qsf=Bydt6*VLg9H{p zpah9k2gKnwj*|dI-&+PGNsq#O>@x%DVB&2qXj<81SlITApeCf>A`A*5<{IC9LhIl{ zhei5%+oR^M$mjB`<}?4;rs8 z)qPiQNCp_?*7M=R0_yW@kTcBVDj0FICP9plttgZg$IWoH^-SiB%D?#tz_JQ5=64VN-@!b# z1%nS1N$UPnWFazR6juKoT~*0?$61{GW|P>M{i&@T-K_d%A|T^ne=5PfB+yn??7wx3 zKc;sV+kN`H%X6E`<&rLGFvE`2{oQQV!f_fGZ%|x&plZ?-p==z5Rx1H!(cfMbCgW2Ah+%lhT;Lqr)mR5Xme%Ul&wU>fTN_KM4 z-*~=MF}cByTB5}>sq*R+yVaLv%KUixI~+Fy-$>VcpN&YFtpl(mcVZN*ouo1{`d79D zi6(@vG9@0}i#Uu+o47Pny?0A58TiBqztZP_Y+}uj^^~(pyKCg@0+EbTE5DcDyy^~y zK=+9It=#`uNO(=9Mx)RZ!t`P{XMpvT0~V=n-&gMQ=DyqLH#)2Ioj8le{(HCE1=FDo zY-Kyj%OcyutZ|9N;6nE3#zSjf^2;dE(Zs)K^O4)qk7U_86(kBS)H(t{H;7vU`G915 z9SsV+UyhxbBBIj^W7&>#&LJGo@EJjOhTi5=LNJftIrb8d&R#{Z)(L8J__=Ss+sv&U z&q4T_!2h;-Bh(>irQw6kp`AntZnUOlnF>y-E>YzYFiK)=b^yGB?*%CrtV92A`JKrU zdNm^2@n{H^M;7aGssp8$A!lk)QjOjnrd}n?)P$xr(qOB7X4j~TDmlTvWQtt_zT}~a zQ5tu)oe)@cY$=pgP-@G&;Fa(G$^Ty&_p4EK@~U7`eY^$B3-u+P@YqVb}eRDs;IT1acHTA_;kb_ z2Fob2XA%qdS&S#w>_IcGF?bzDD0l$Oo(gh5TQhE5Ofj(?JiNLSD#EY;FE~Q7A;Vl3 ziT`shv_ll$`tc^bSSw;2JU|C8W!Ui_PXnt1cII<{)P6m(uFJ!eaWj86h3)J`CbRNF z3GK{af5LWvqR{ba65-mTz?!yCgQnu1F$>RqkuHIY)ClgVf<|f@TIHm~THaKJNWG@3 z!+Uw6X=1R@7Pj0S9`)u{M`eI4fMCEQDi!z9j~r-*Xy#B{nF4MRXlfKq=i+A8P2>z+ z5=jJh#W;cLTXj{7jf_wSHT7&TLIbK2ukW&f*4%kf1INS{eGn95KaH=pUq9NF7rN>Zok2fI zH0|3GZC9YBwtrngbLrygj(knuw~IWGb=2r9H_5MH7ghingn=$rjN4&U60B`5C){-GcF zO0Oo-=IS_;$~j2B=(qBlBPL<{W%m|LF&ZpY-JEKQPTn#qmW~-S;OUk{u|5Xv1@wxX z&g!_{&q4~yk74thu6QLJRTo#F9HZSBIj<{AJS=8Qfmv8x(MJnEprxXTLAvSc5+Z$a zT6z9qp}XjR+DJVu2((Uzp3m9BoENA2vgo1QfAf{=v@F$h`-8R=3bLzO&ttdUNPxNE z8TlT?$*b@wY)W&Jw@!v)g_T%a5X+at|vP zo_*F}6ja-C*lGcrV`)v`^Q~%D+#_4Nx+6dNecDY`R>LOh3ZrY%h%Jb9FWesQ%1h>y z*!>2tUDGF6xq151EP_OjEUOqIy)er#Gx0-geFLpw@Zo}ge-rm{FlcvxKF>t(YJIwT zJjEM-!a=kFK1enH)(PP6v7{%O{K5#o$6#4IB*SC`+3PYut5lrB14e1zul7Lted#H| z@I7G(3wzc`U-i&s-_+&I7mfn=kJ?`u53Zgc@>(95c3i!S_zUZ7eWQ)8PAYj0qJt>Q zlmgNa!2pBtpp@SOp*h`Cs7Ng`^^n!-E6~ShG}0|2U7g_gtUm*HE9Yd+Cb;w{2c=0# zZ4$OTHm&S!$J25mKIKaT4^No-YpN66B0X~7fnkwm+Q3`=D){zycj}cp90hpn#_=EY zlw}`RhRjPf53L;JZA0xC$a8f~o0ybCbc%5C~P*oBK zJ%tDyzUitNhlNcZk)*lrB&%=lJbE+_qbB3*Hs$nTYGe!7bbJZSoVzbkn4G|GuEqcT zgY#jTTR+=fLWjzn59l9o82Y^O(qjmp{y91)sOF4;Q1G}OW(yED3jnf2*QUmC3c&-ZwPh<{&qsZ^lG6{4A$) z2jtt_)GBj+`CUVgRK(4H=0_&hhx2wX5Wjphv5RMT<4TulptMZ~VYJG`WBX*%_l z@WhoQ)j_&@8W7AAO_9z91pCJ&={r|3&eY%#*FXJ_0w64{Y1+9Q4Hc!?J)yuLMvBpU4xsx$-Ioh)46iW{IjiPrAKxps)Jix;~~$pF@QNuNSq-9re3ojkmL*CU+a ztCn>igDtOYxk^uwVlECH7Qj;ci7L#hV2&wavHfEnK4Z}DA-P4_E1XEUtnw~_ov)c96?U- z=7?^|YJY{gq9acR*Y!utlGLryM9SOHNIRC;iF8_pL;`0ZPh*0Le*=71Jy8OgauUtF z=I)JqGmVR&TF0$(I7;sDji2S|S&Nwx$k|d!)SQ7Aw35`6BW#nBZ zQOay;1z22#?`4J3g}llleEI(Mm5Y`nBrkWQ@z{hhQ&nwB|6ru1N<>H}6wrnW8^W4iQ#2>={E7eTgL^t z`EO1||44M9KdhWUa`>7qQ95#U&wJZ`tXDiUeuqEMJ5*sEgG72Zn$E?obFA*-U`?JA zLW8CSs{xXm{>MsX+YDI{Q(DL~3ne789Zs($S>oIQr)MspHzxe;$132$OK(b{wmee0 z&iyrqdr!A9Mw~MxXS&gwHNEn%;rT_z5DEB&G@-zY)YierPtDjMi9r`J}FUOAgHMSpv0}kaazqB<^3EGso~8j6dD)H=ClJ1cu9m(P;Z)5!0Rx%*9z=5 zf%`;B9YJ`fD}@+b#eCj@c-TkdS)LMFCjkc;Vu+~RCY(Z&Ck~v=SS&s|@^sEPYv}AK zDZTIZE9quQeQ=^;L0gRlxAZupbx>FXNB3IqTn-j_#t?M{9)}gv1&xi=53}p-XCwqM zT{nb79Z-t=rI7nS+CmwN@4q1To00@PP)Bu;RQ_9T;_f9Ut0kc&ANX#%y%;lX`GUEx ztvZ=WYf0J?IRb2A0(LK4ri8bN3Xyh+u9Fax!;wA&}l+SBgnYUW}lpw9AP+^%xMj)&;f`0F}CJcNFEXoa|$F|~}2qjRod zfBt-G18gjWUe!`@sYQlmQh;80TWDZl&ESEUYfMS57Rp2G+ot1%ivFJmr%@u{F1~2}Y zN>cst5y>;iqO z&r=sY)A7Tsn$^|96OZfWtt9phD}*@UFi<$p#9Pp;JS*hVS68RNfacTv-Zh|0bFf0$ zkOvzZ2_Vd#-4*&%$;j@JM8P0MEkTJfzq27+;tuh2`6Oh1U|c0FBQNseLpc zsB37re|m_@4bk`L;wq=ND(;#NIj&ZI$4OqKg@(e-tB_6rvTEyTEtH}|zC@0JTqIE3 zdSv!V5@J4k|dPrQ>;mO%P z-X(o(9ovJ9<9uoKC^yb^oYF5O0y{dZPTEYoQt&+{%U+=dDiL-_XGYGaz*}N`a6_Fu zo73jEBD|I5?TBF-ljv`qGzvL8G%>Vtp)*)k&=tA3>ED{jmpFT|XSPFR64WP*Z0bC;|NT_u(Nvo) z=(;@QBt$nS<)h{xoxBh*6b~40#%-&@k3jPsv5!J%J;YspaVET|23c??>PA1HG|#BYUa0OxKu5-yVzInB5izR&mp+&g_N_Z5_vyIMeheQY6#=bb{cjlE0AV#n%{ zK(XkmgdD}SJNvUH?P!JF%xY~P>!B@23T}O02j`+Y!wj`@a^f?2O>90g; ziVL^jNLF~F9K68OgwvMgp;`^Qfq*d&v+2R`op4C7P;?t;FpOh!l>FNl(K`F|$*)Vn zdor|T>?kjID_G=ilxtJqy@cIe7c(-`BY8sg#uZ%~%6)ApQdEu*+X9E|(!PeMhz&4O zL>V(BgsgpCN;PNrd*}G!B_dBXy3#R~v5DBnGN(Ex788&Gfq>Ia164w=$jw@gcd z#58rq1;otd$A7wU{f#4ySPh<;AR_$_^0(Dg*hVf&?}!PS`A4WPut_PEla~*{wk%iL zC*yrgT6dd}SKW#8!;7J>_ZUk=LfvLg09(6QYL9UxT=jFkoj>~a?t5w)#99V6T9V+8 zOP+r9NL50=q_&k7!sVF5mM7o+eOW$m&9M8}J7ww)_x-#X+=}^>6hgJ*tesdS7?a6L z%b?SGi1@IL$FBt9FVwt`D}C0gXA`dAH`pyQii4y9U_xbR61@Pw>=?>;vSik-3uF2r z!o99Zju|!qTjs*89mly-yLiF9Mt1LrwXppx{u-qbTA!y8x#sssnR!^(u8K?}GgMF3 zoeFaLPt@%qEOFkU%eN%MjuUkEv;GjvuF7>bJ(-IzBWb@Vr@{B+2#2+1lLT+ej$3999ltASaqJ{n2ej9w7PY8tE)TQe&*Huk)Tyvdq{4i z;kuv^$=!A$CqLngK~Gq=Ul9UyZphH@aIr$Aw1mc2A7Qtbu|DHiu1Ach^-BCi9So0; z4dR?oIV^e#}CPFLg{J2H>cjL^@%BL;G8`N^bS& z&WYJMR^dT7d7xf6xyq10RC?0N1BLT6uMMA?k9Dqig7wsyHmQr9KXde?UC!G(0`({B zU$#@GWW<4OYfwpVt&bKyhpwE|332tb^Zj6e%6`B$XbK zQ=?06nVIDOyoz&1EfaUg(`>6uyidw`S!Ij@mI}!mF6VwzM+wz?qMzuI-PH=?g#d{H zNb(I7Wv-wy^1^%N*Q7{cI8>*Uf7Dkp!1=A|LYtmBs%sgYD1@-IY~7K;9Ww88I@%$d zjq!B&=BS6#(?b(4S)+_Pe1FjtCO@Wf0iHf#!AXheI^>Sk{iSI_7nFSkKi3yR9jLW6 z!Iq1sTU5&gvHBHfoY{wg+$mAnU05c8GVPKoTy1+V`{}C@vSjvgz>eWR2Tq+J01BYe z0%`pbmLC_&a1f1iepPf*DQW2?abMMS2T%tZ{ENe1S||c;ny&0wAt?YpZ!YhGMrG+2 zD4G;I%Gqm=s?b+v;#7vd?V)(di3DY7d2J%_3zk$lg7lmj8}E}Qt5mtC8X7#}&I#y= zlArVz0A?pnYHm#*?=l{oVBHi#+C;8aRxEo*I4_8-;*DZ|jRSo8P@6bd=evqW9z*4) zyp~yh05AM;SszFfO+>&0p%(lo66{p6j*5t7SJrC65+9r9zRUWi%b6V4mXKCNBGCyN z*2+o!LBttINcY{?(6!+}qJcWMkPgo-dc79WjrxJ+2Gde7;i}4&mgbvoDctPOU$P#Mw5d z+v-w?zt_su6|OdKGtovrDTM!ghO!`>oWxiK?#%(z@s=tCW8JO9Q$Avg$05Sgw~SlU zTx=#Va<(nG;(l>2Z1*CZOyz@Zoj|OjHx^WOJy90H=2osC)ZJb8eyT1P<~!vr6K_iB z?S++pSX`MiCgEZ3t(&~a1tnQ(jrdO{O{9*rpB*^wgH^vWB#!`r@P{Ba=-3CgjFv;s zmvBXF4c9P_U6H44F8TR(gS4bgp+pmJN4MBGp*MR*_PU9_JgE z`6bR!mP1Gm>WKFvvm&&hhH%XE<7kaN0~`j(4PttBhMdQ3Yn>PpP=ZKt5O6~O8aIBekeI|hZrCp=a_8V64GGQbBOME483Q(J4}Do4xUN1 zq-xMWmMIV!abo#R9*3_em}Z{Uy4Go~NBYjVI3OQ!EhhHZ#sfnUMc7zNA|8ivDPamj zO7%L)YMdSDaJN^FsfS)S43*X;^?>0?^uzNVoEq^?vhKZ;9DY?Q#4lHzN`BOo@bTI) ztGnxr2%^p}4Pa&l3?Vm=a|_Of{f_hJ9wQ$Jx+o>&p3kf3a#qORif1?*Uz<$d{$?WW z7$Rz*PTS1FiDMKuUJV6?*K6=&iEHDV&9nvy*;!TxWotHc>nNL}(XTXUPXxU@wqT?8hV z4lLW;OZw=c)26Jk{RAoW?Rp8xH@hB7d!;$PZ$kE+=m0;)q(lxU{Pf`hd0*voF;f@= z@p>$Fsk)#_WL^CgN^R8Nqm2%n3D0)J}hO!Hf@tjqbf&>R%ZPe}4L-+5i(0+#f0efYzFyfEX*lVZjz_sQsmp z5)SF6!=SnxMHSKC>Nd2sviPW>0ALh<2R5U_ZG;PhHKExMhGYglU}i(1^3t4{*3~#i zP@BHDQp;sP9ODqw5O;+d}f;u14}-Gs@9;#_J;r z^p{No^H9i;hNFjq(b1h@4Uq>(GY=XG1ChMi3+?doWflo7U{ZEIZtR7BtOPP&@t`_(a^f^0|t>x+lNRfjJg5$KksMA|ynVhz}aer)uFH)j%gzoodhR z!A@S*@?GU_OMgf4TcunlQ@%lhNZf$>?|9i=?k8rGEo4d7V=@$&5PJ;;4^0WFSL9!m9?*J#a=-lt0K9M05D=HPj#aftopgj#P_ z8CvGOGv-lUH3Pj`%9KQLi7X%mu8$HCqe7RBANT+tp0UJ4J$A|m)jw(X0hchP1yAQ)Eay`MB^qc-9SvE*J!3wH>|dL< zJo6fZyS8q-`zaDm4(n6*4iEClEirkJGQH)>ZTNZZBlxVTuF^!}w*WX{vnAe{(v9b{ z8^!))or$9FJ!J|RBm_=ZyQJFkvbu%(J{L9r>znHQVmj~uC2hn-(GZ;Wxhi|G49=7( z4nXUkKx?Fj&A~H)WTf>sYxA!UG*%sN{85tFt5(T zt5tQ!7ktkKnE=DfqPR+()z*&)XRX2Be#oGu^KoT#%r`0{)smzePjrJW<1- z1sp~kJndh2Hy=s$IAk#4WO&o>`}s-*_b_R+BQ*kum{)#;hjPRY3_KtJl)ediRubF2 z{dIcCkdf+_1ei-<&4Y*8>LCd5VE{HygZp{xG_M{m#jh;-C1Iab4_o{f#~2g9GeUb% zze02)s7lQ_#TKm#u3IvEdZ|Nu^Ax6ZN#cH_gm2xMQRjcpT2UgX{W#k=a&?Z3zJ>S0 zm3R?Cq;I^Aj7U>JhJ@RnfcQfx$v@p_;Y&sAY-HPIs}-c8Z%$LjCF@V58sp-CNOn2j z{QqqpmqU89Y2~)Ge%u?b`;wlD*M%Hna4^Q+BnR+e-XZ}BVra!(KeDO6W&$s*lwe$^ zo4MqD0+HE*C4D=t#}d+_KCIW<2zLt|Bi3Lef)S)FKTKwQzR??ut50QM|!; zr^V|aY9N`!b|Qwhtq_fqg-NUUI$dvZZk<5K^mum>gn|s_o|WR-=IuMS%CxeY$wzz* zZpI?-guX@9gnmx-nfu|{T8lZdjurUFA<#Bt7J0osNL=O=d@u#vLRs2!r~)6m=J1iQ z$w!X9w(?naY=5y8O8Vun*nF5@x*H5C#g<~^a#omFn#+|s&Sh2+&ai>~HL)=wlQS4D zTKBRoC%qoMDT-8H$OB=i$e8SnEq3$&qnpCAL0yjBBWVBFY((G%v8jTYWN}Z=lu@D$ z$UTh$R`7PeOi|WUpu*QbxYBXJ67?P3G@FW_)X zOc(;x{ICOFPT8#l*K+kDnkncWF`kSn8r5MTFyWy5LqNL@d(&$%M}d(koNuXa=*j7l zpvM)pdZ$P58&N2;i=$T!b%X7$wG}9j@_#4;ty&g)XapkR)eWm3j9fCw5}{tjWR%r$ zTD7GWfYhL%6XTQoN49{rXAFxrkVLzMpX~X@ioWoM%bN$pwf`6>SzR`#a+K!KitHwj z1p90l=fT3}JNS{~|In^{rH1}L=Pz%+2W}2khftE`E%T9oVY(wmSeX88IzJMID+c5g znPx(=DQa)7TZ*%QK6GJx%=KSpl4{m(UXK~ZMCVLC4PQAL3>;aST)HX+)ES}``p(VJ1yr5T-nDLw z!a*;9(eZ{u3Ohf*(zo}R+Y;gcu2gr+ngHSKmG#rsfJt)`1^9Yy&Zu_wThWch0|Q!c zhZB61)q3#ncwv7@CN^P~-`vKBBR!zyBOs+Af<Q|ydnm2iJi z>G#3mbm@KAp!fygJV)eznEvgNd!k@qI*?N}6i?Mtm|wm+8_Ct;9IY$;mqD+$04d0t|Op#A(lguow5#KWtef?YC-(2rd& zz`P#i_#?tcu^+jiR7FxOaumUoVfIyvAR<>>m29Z*O?W=c1B=7l*%ZT2(s!w7O*bC8l-z@hf#UcLkEP7o$S&bfT<#?gh=^>cpSYo(Km;zx!Tc@gif1 z5Xdp+S%RyfdzB^HC$+!he@gB54$dQP3-~D(_y*pgCGV`Dh=9eiF+$<T?1iN$JrFlkn#-tTBx@YV=}7XXJ^uw>F-RwhX(^?5BdlIVTlRPE8<|% z)`UnyP-nPUFxI&Z(p)3i*eQQVM!2tIHZ|uuHg1S#a@n3bIK9>*d2Ft|z7P(!gFU9@ zDJ*u!Ul>jq7x|OTP(7XSeap`7Ax}n8+Ac6q+HWD}VkiJ|ju&UiT9qy+#4FmPyB`zi zOd7j)I-I@|<`AjSj_#dvUg8W^B_YUC7XL}GUel|5+p0r=raN2erB!x=+)XMcS-deF zAmH=IHkKAj3*`N9C%E!+Ndkl|fo(Rs!3*UK4&`Yd#DG9~bOqV7xr|-bqg^RZ|KBky z4=dA@=3E~>C@qvHd>_VjbF>Cl&8^?$T@wC-vOkGTTb_>tE_fPr$@8)} z?iXao(az%F&}=M7jh3e(NyPmP7Hb|4#OZnlw|F-B+$Rj%ftC{{E(RH$_kZ;i!}f6~ zwQOV3Ct~LCOL$k2odd`_QeuU1tdh+$rXy4<^LJ1PPE3~r@?o~X;MULdK4?DPScil0 zW9>`5X*D`2lapSs_o1da*%;tuMwvqFI_63MYxHIXKZF4U-Pb+I%_qHM*iqrtLktgG zV&#@JR4x}xehUKJGq^u0>DRpYAS46kD(Q{8ESaNx+yNM6rKn=Le0l#vFD}N zp?qI?iel;#Zqtz~OHCF?_CD)bVY*4aqJF7iu;v24Dd1TP`BdvMl77$yyGkEC>FUO{ zYa%8%p3o&G`w3*qgUtFMUL>(ssKB$`QAT*Cl{}{7gkDD9$l8sxJt2a5?`hq-6UGa^ zS=u*|9*^McN$Zh4zII^cnn9%5Ew@O($Y~vdk_b=4Nzlct(SPGk3pPE>T(;t%i4DF| zyuVY>$O;QT)hb2;L;UpJJX)i~`P`o(72xZFDJA-K+r_TTUcC8Xe;V4Tn)PyCu^95W z#!v&S-AexjwO03;%l@BeS4Pdv*ujJyBq3Jhru*EOKPm8zhfowNjwJ}V2+G_{d$qmM zth8QYhLwi)=)0>63TarRjWWa_>Lr(q_XF=aC&U3xvjeee$V9(mu_}SJ@D~Cu|F{AIjRyr|cX83`)ddcUxR#D&CAI%s zi+Xenty7iaC4m7B=W*ptH#)XnpgDy|K@f?{KgAe&?re5c>6_j7YDsHJsC0`vA+@D7 zgLItSsnms$vdTJc=0nbe)Ebjlt zFC*dP2NoeY)VT@0%_M5C=I1u7gO0SL!);A&MCf=zAv-h4r|N}kMdAm+9U3#)^tU#C zEKxT6oE^ZlKq8)41O{XEw0moQU;!n99NF2G=x}ua!3Fv!cM8IkKXAUe3g6Y#L*;=4 zkc}|?(BVy($YImt{9=Xf%HJamk;v>9%A+IyPMgT*z;4l6gTK(lZ*s;3%ev`{b~)bW zYk^#=9=9YrEDbHSf5G4BOs4b=p+LnGM$n19o4JY-)3TE8dfyHmzB*b}Vq{$O0zJtw z(z#|obs59|fi}hr3MEeDJu%+b(_%oEA0VVQC{FWL)nJzo|yD#T>mR zUVtT|1()JWFhvxIL6mnntKVH&P0@j=6TK#am+)!ij+1W|U-hiFm&=KTA>Kvh{>_Ps zqa3FRnKZ9_%yqpqJsv&)s1o9Q`ISFS;__)j4k> zc6JkVZXbtXa}~E;@}DCl{kT^G6J;(FEPm=Kh8SVw0@avzocdDl@r#^pn3^_d<2Ef% zSo#8*Xmg^BeHH@0_|AHbPj)3cJMnLUP&kGAy~==vithPXYMz*96M^UYF#iTmuDhqR|J-LW_G(cOMyMeUH!r@vr(EO1EYXJ+}7D_4O7du4Ha3f*}S~UOzZV z*v9%?Ub6ivMRRtemrrPyz2i)o@zTvo`dSuYkLainJJTtplW7|)Yrbf-Zgj@{o;C~> zZ_a1z_(VQHKJGijY93?50G?Ia&#;lY2|3wrfd zeAHMbZnqhf<3Cb))gNk59Mm*vjstD^sSEa6oVHG~>D0+E?cmkQi(FCI)ep1pojr*YX zJ)Y)yEvzn}JRE3(KvB23Y4-NwoOOcPE?V-}JFEEMACmCf@>9u3LrwZwf-A(ehfO>- zalIQoTzf(K>Kb26fKXsJO#-n>H@2_ps}}dZ3T8fti3U&?X&syG2Pzgds9Mw2ndD;} zwQxY<Qk5;Vn+jd46Ro zFbkT1USi+v#)e642qxAaO)oxsxD;F!S5d#n)<1EANa|#V7oGvWgp3s+hDT21U08SN zC1{j%Q~A=rXP>W4v2-KC>ug_63aRsMQ5#9j*rl%y|0cgpu^!*~XqRs|<1a-25U{^@ z(3Od|hRed?%JaSWx~O(C& z7TBQ7Zz7_@r$$_q?6rt|tAjkBCM5t=e{wi}Owp6I!3`g+>T{mq7kzI`cqQAHh93H% zoq`cY)huDEbI&1{`F!i>jJkM`3kHKpQ4*9_UPfo{vAuCGAoGc`^}@{bHB9`iW_Mvh zS-n#A$~H%`0AsvXm8O-kqbd??`*ok+L}f+Ux7&Hcp8k~fl|!YoUw|{5G_5M5`ovUU z0V4bjYk)Ik?I%@bo!+<6dwm1J1;$!U#+`*3j6T?(y|?H1t;gu?w>@~cbwE0yZ@{iN zzY0!Z4FErCyi-UK1ue8(GSwjhrb~a^!efjN?Qc|ju!ii9h^miB8Aiouzq7{po{{MP^!{qUt}C0J)GM54B= zb-!&ADlZldWkh$`&Y9o3Xib-xwOmKDNOjP@&A}9M#maF z)^T6@-ZW`aUSGx!lk;ymP+=V}nrJizwFtRd;vPZ%SrI#kCZ-p^VUACY^7-MXYHG}- z1N}<GWIzKY_~$#WA0PC zGnsCRB!legbt#+SR$yjFg29^hzrVOqUWAi48SOtL7MNHsKhyJ;)sEsK3*ru62TpbF`u~EmPMGiHtN5$ynTb{ zHEguF-V?ZOa5A;)=p7R9r^!*9a&Gp32Pe{9%ZNqlc4y&{whFiUrBNlN=m#qUt|ai~ z!m}u6=NmUO86#0QWupHClL{@7-@|W7V2f8w>5A(ykh4k3$CekN_7G*2!vC`R6P>!SL!3W(fnMO zdXg`=!ppF>TKT2Qu*q3ESBf)5DfCN*Cf(0YF;%7nGpzHVLP4|lQKHtzgln8~gJ#&t zBF(D~Zf_W8A@z-!ZYprFA9ZcjS|Q!)<&@DF(O3%zcH)+xJH8E_EZ1K^^y2G={AHbL zFtEUz?2U(*=Q4;u*8q8)fh2ac5(dLyEk#GLp)ddRR1I2N1EF>K=2e0Ue$k$b4hegd zPhNj=V)DuI_qd(A1Tz)c17IB}M*EyKo(-p^Jv(HWwUV%nh~!wF#k zOd0xlOft1;5(M2x?FPZm%Gy`Gnh&?lPomL8sb`4K;8-K)sJ zN!DLU$Xoz|RPFq*F*%LK_t$qy5XRx)QdQku*wTm=H>1ujat1-qLNUO93v%*0Yg%U7v6&nbNb7>{7u#lg>NY;N_~#Q0|5@qcQ4{N} zNvRUtuV{`%%Gqr%eD8nY&axq_Eg#dDn`B{7{MKxC`1ot2#(Stx=4A92pzq#k;9}g`tXNwu z3M+P~f5X(cTWl|{UnVGxTz7{N4c$#re0d+;%;-Hl5%bt+XRulX&3!jX&>K2o(i#42 z{|k?7dwB}wOYk@b|FHe#*g2gho(CJ6E^Z|H5APRI?P{}PC`{h8Iv+e59ZXM$cbKle zA%!nfnAm?lV$g?GR8Ey5GkiJB{g%Y0o6~BFVukdKjDrM&B(qTo4>x*s^8@C9 z(33Kr)(&a9D6Hv8Fc9yPU*zTG?n)Zp6NlOF2|}ed$L)ap31mZ$WZno*73~nj=JAaL z$k7ZkqQs)9Qr4mlt-DU;|G)ptN?oIw0eEm7CX`6#2)Y}JH6y-rbPmMy4k;3;eK#t( zg?)ZvS06J<9?81xJ3M8594%4^t0XV}2iy*wFvxCAU9+mQ`MrB!|C_c*ZqoMDsIN59 zIQ$}Ab~sLb-_JYwx8`G4skbibNDX5gb!>8~SqoMqi*Ni@FV5_4FxqY69UQeWm~<3G zh!uJS8`6Wt*%4q*fki4j)VzOOq|hf?H5I9M%*$bpPnV@2Yp&{670*C)R+a0hg62EP zAR0{~Y2H^-@{2Bsgw0Q*=GfOX%RqDG0W7;XnVt|2j>+g?Wm$knFJ><)(8jz|aqY() zRm;xy)Sa)u2YYMI!7QGyO&KK=jJKi!<@qa1x5&_rntooiS&N-Qq?maD{qZU$V-rrd z2QRMDrMuI)miv0o5YbDL#kWacG6M#^S<&mVBu`UijHxoElE|1J?6_!CUsV#lL4_k( zhEsY3r0SPYrWcbYstBWb#P~@thlM~QpbbuVlEwN(dhxx)jdWE`Qap7P427cC=M%rVLAsUo}%QvTX2+Tb8n{`75^ zOhlBB5g!LA7++m6*DHbFTM=F3Ci9Me`hC<_aJwFzcYXb$+R(1^BeOAZ06x2qqecg4|F;;VvU^>TXkrc@wg!z%V}{55*fm zZ^MmO$8x}x*j8VH2{+3`D@#CeH`rhgfGz|Z7|eIN$g=0#ozNCaip`m=G7&Q7wHm3Y z27BTtW@R6oK8D#nO+RX35(H8f=mnd45qY&5%f5}g2(dr?bWtb7fj=FJVgSm$L0eRn zNABm6*mI_Hs!d2aCbQ_L;LD3|9=f)JC$a*_$bJH8in>1DX$N0gLODj2Sc7sW@v*0N z@+(O2Q=!AVvPoC{lN3;MzJ0&}Gkg5*l$u8wMO7vj%>b&sj<7Q-2hE1Q(Qg(q+259) zv4W}ajhFi=pY+itLLdI~*i1Sy^q9wqc-3mq*-YjWm{*5Q455De2gA3=K7i+=#qo63 zdKzO2w3qU<`Q|qiNsfCD*ali%$ai*8dETX~RjqDH%B?uvkFYS?BK*;M-5Mz%P|lOR z8yi(|;N%HNif|tBWgRmxu9VfHQxq|&$yPn4u=zv`Cl7D!sz+-$G_@M}6dZ}hxxwW? zxrR#a%i}P-o7Jkfu6h6f*vkM2yAA8#DByfq9Ao}7KNL|9Z{F5;Fjvvtq1hRz*n`O* zACYF`#P^vGOf6jzOO;ApcdWC`-CbPm?C~heDd)>2z!a|{G^x)H@bB5t!l{*YrfzWg zkXnZ(PErcV^;SXf@<7t)f*#B9A0m|F!mybwP3&1^Y*$F19K8?cIKP*S$I#UP>N6*k zK;G)FekK(_|1>mPTA!{h@C$CD#a4L~m_QeayRfa)2evN`zzO~cDQH=vQ4k?Tg7{)2 zGS|7ANj@qG#m`07@kdcjb3!->;gFg7RuE4Cx@#CJ0eMNtXu0N;bZG?CquTVS*z!Pb z?{;{8&(*N}hac8RHzr-yvUP8B2{Vkp2%=hw1;4UNZ%C~^V{}pG`Vc)$6FZRU!^x&I z>V`~qfU;LdtFSoF5g z!==5Hx`*U!M;SpBRAgE-Itr&5_U&zGAf%$QfBo~l^v!?_k#hgF+5iis`PjUUc$w|A zS~etD1r#jYYeq%cX!OQ)=vhJde$Q9eHI+B=2!qzE#D0}O9>>fH-?Vt&(XZPV|7v@* z6Gyg)O_?bXwCD6Mg{#05OXda4{BFoNA{L*iXPJYea!W?P$@j3}{oo*pM)1($_^6IZ zPmC=qQW>Sz+5)+3M)*+)rn`-5dJ-kFDb(ybi!JN--hacRL5z)tP5!y9-?dDyh~og5 z?n}<86bCLq+8+y0bfmVyO?VYtnnAg4nQZl2;!* zB%CVS%tsjw*J6x4EP2-zYOH~ONAwwqBnVHvgXSL1^5za$T~dRC7A*myj>m~C8mmV0R0|2LVazV=X}qM_+<0hGTt$KTx(Mpc`Z}@s-A#c05R zkr4c>fmhOmW7J1rr1#=~LQ27f+7{Otyv5{|V8T=uh_sPk-c{=0hm*=7-01>E#7Rh} zAah+{zhmxWp)Jdsb?7Ii$T&rA!>(nMxHMKlL%+M%S)xdAwm7CVVa2BeQVWo^%I8mmS zENqqh#tA0lX&hav(NU~R7LNqPZ;aQD6?~7%|Jh!M?K0Ef6$mR^buDX9M>P`0oD8ym zWw}HUyzB>s!R(9 z5<+Q~%5#v83R2pM_^YH2^1Qxa4PE8}GJdDOhMfB-X*e6UO|)#fsTnoIJ7^H(p+aM& zPPj8tv6Rn{DLWSU8mzPX*Dt7IiT-}=5e2wI-ZgQ}zhJd}adOkY)kY`n~(5v9$I z?BL5v$C_{fi%CJ1`)}TUc-FqUWNNoYLdrQqZuQwij8^w9Q(>{yRI7YDERRUOXxD{LI?5h#VU|}+q6G(blTt#~f&=;3%V28(AE;Yc@VWzk z5-habZ;SM)QbMi(z7_rxDI*I>YP(h8OK=)NvY@*$=?!}0;CjFsA4V=}bH?zW^m^i! z^V?(3%QaW9+xCw!W9(Cx#1*l!ky;Cqr4GFi3*ir1l|u6vBMCFB2u{T##u@3p{a?k` z-dC%cBfE`5!4hiM&vGd)za*vRcl|Kgcgy8FJZ&9NW^o@d^{9=H!`%YOPf13X@r7ND zu5uRtn;uji)VCC0T(+AHS~@RPp-h3LOVTT3T5@Q6rQD)T=q4vQ|Gma+fG{(VPD7e1 zWNy(xcOKLd^a9S(V84f?#+(M5PWTf=J^QtUTZ}hZP*fg_=5F0aE>!%zTsAJ`0AehE zQ-1lCjf|%KpBY7?Z1aSNyr7|JGP3S(98snB{;!=wrj;DnC}DU%u(eJkZTZhP^qpvw z$4UT~9g08Dc_%pcUe~tb2 zOgSX^f8iIXn8}~?@s5L*QKAl30(&lH306oNgIm{(X==-~rkNyQpCMTxwi)?DtHhj0*3vGW4g9B<(4i1Ci@l0;`v||?1GV%c){0he4^vy zw6o^Z^48d-PW;)`t3VD9AO~fqSGzS6KR}e)QQ*V8W2`BC)=BPZsn8~@&P{sJr&*35 zf2oO`dk{N1W!N`HS%SAs?Qp+m0A)oX#@`_rJ}GY6l>-oMp=>m}IQLdT3BT{eRIlL) zg*7Q>%K_+tbexnv@J{|}2c-c#ASQ$I8QsoUDSBiEYg{D3`tnz3o1;0clzI>GFHco{ z)J|p~e3XFMPJhS5x`8V5I|k_`p*mQ?YM$cUsjKu|i~+^H!XdOP#DPslK+JoPR@%3`Z3Eu&8cZ^Q?CH zN2>hf(c=J#ZvEwS9JvipTpG5fYa3q(lJMdF)kCLJKW2k%?S9|`F6Icm2U9CNGV7R? zEY95a;T+W|pg_R`53D-682BJheoTItRdxwwZvi2HSwwa_n6LpvOO&V_qZVrxM8yi;#ULDfw4tNSKFbcKemkO3vf&BOMy}RHGG4$N?7`zhMvt zMMEA~uSNbZYv?4QvNE_!GPiS%_K&+9STkG4#C$h?*WD48I0S86eSdF}R`f=g)8KAK zDJzU#R>4DxM$nd#GX#P76rU*NF$Fpe9J#wGy4mbyaF1g}8q_5O3a(0_2SIQOgM?a-u-Hx8* zYj**3AXr$m{d$|9MB&U4id{}L!~ybTmtecsf-AJZmr;&#(mt>GytxkKUd*P|L+L?|)xmb>EBQ?X9_!|NL<^&K)L#JTZ ztHI>Q=IMiX%m54KEfDqp6A0^nVGC;pF<|7fwV)B36h(&;3M!~1$)yW+xh?Fz+AIo& z@H6#paY~6^?pkgiyW70)sEZq8qG#gYA%mShfZ~|h43jhKcURpp_rWCGHvVHu7jl5{B(&)~5xCb#xW$)eLkUZgOH zUkwxbH(_cp8p7I?TD0%0_I(C@P#KEPkLSA7W|}t4yC?B<(VS zs4u4oUbl2%#xxy~_&_%v3&S1!Lc7%!Huo9SC)Q{KF1iAqq2uuLrZPmX-z7R`En00? zcjmG6q=Xo4ZP_I1jHy8cEixcLIYo9{J*F;{^SFazn)gLtRJ{TMt&lgHH5LK9(5Que zRtZeQZVo=;7+b5^56pZaZHH}$A(KZotN)zZ633zr;U)1QpW<5vP*%XISLsLfrWD*i zd96=M&R0C=is>JyN_@b7sjjR9B9$WS1jBf!Dn6X9mH)E^Tv_rAC6T|jy#+WU2dMw^ zs@rBIcL2g-Sb`aHKYcdtvdYL6o7}qwp$&vk0bEPO9LGnVhINfWjM#)L#jU3Ah4l&> zwY&l~g14rt)6>Z>dZAt%qO0x4>~sBxwqP6^Aj?nZEQbapaboI>qb_vx|JZxUgxQO5 zTMGg-XgnXdoXFM?-pgZVRap*upevGn$A;yCfd2MCEFeCa))A6iP~TIyY@ZQo5p?vV z>@L;;E}dKjRWR7qKQXa`@j5t9T;luKf4iZm`tl>nZs6SbqW>*3Hi`gRoCT~C#ZE9-RIDWTBf;Y|)h^N|S$fZ8mJt3;& zm3N#}Z4a`}&3UBDK_Z}%)&6urk>IbD?Q0qoc(s|!GJn3KLHeI;tC^%KjtT~oWIL{mqh)unUF}bUZ>N*Qhcp1w*0>wJ zUz5as-0kSo6I=?S&bC8#IC~4!SB@WGUYK#8CDNW#zyUFXoX4l>A063u4zjtwlT>R_ z;yV${SyJlG86bH~J_WJpocOj4g$$?2)0Ie^aNA@MBS+$7Uj{lCp&9T+j^?aAT*DcMVJL2J@iQ~>fqM~?j?$BGeP2@_e(zW{rzbBm<^(%{Cn z40W_QDa9E4EU2}^&s$mI%U$U``%tMbj+iv^+yDuh|HYwxEyi$XJ^Yut`!I3oWC|R_ z9yTB;;x3jp1F8}3IUS9v^$e(q{8@h7iqWu!5 zk|?RIhaT-k59#JXql9g(v6G}6S%_mcK`=Z8Tj1x-a9{2Xqh0JIm#h*CAT1Ql#7Fmo zEbW;#wtb_~+cFS(Cp*Y^C1bvfSihN9XKcac>BldMA_<3%OecXOf`m5=3Rwju5vCv3 zmBGQ~vgqd|-eXSLY^$K(qPqe^E z-yyOss8a4G`t6%Sn>iz_=q6r9-%ygQMP+rci&G7tG7D>F^tEVRUEMZ_f4poSln>AQJz+6b)egSw}!Y!mnJpe+TW7vcQ}ahJ24RJh%n@@1y4jzLySUd~i*d|lcS36;VF-U6&kWX`Cn6IFSTyZF(GqX3P zy>OAizBrQI*RnL@`;vH3TH8i?ef}EGIX-bL*Lo2StmR*wP4UFhx65qJRMvwM2DN*X zRc~U`xB=sD4JuA>Dj?@SdKjjx$LHJXAD7vjzopw}ecTlA>Z-bn{6NX4z^{VpYt;#| zOKAC}>dD;I7KANy6$o=%0q2DE)JTAXpxvLlJGM8ia}X)T8i-vbo*=YPDdr1`qu^|$ zrdCbb?hmSZJ>p^*FZ+pxw&Jgy+^5n=`ro~5x`L;F4k;`!ianm zEEhfwX=Z=?_HSLdV71&6;4Mtu90Uq&U)b~zJ19q9<}3BF@4{&v8<52UQWdfwLz-s) z6p&;!8-&hiI!}USlWqAkC>ZE9j}uI-Yh2!kP2cB zcGu+A`A8WM14Yk#4PKLk@ya=6eiEvp=l73F=w!zYzQ_!98Lm3s(Pg1j}+49EvIX`==&N{D=JzcM4x5gH%iS5ws1aHVn@9+T$d?piaH#h?gqCL;O!)Imp7 z50LZfmn}1+!O4VK!RWtLZagl0Y$R?eb0NwBUmDYJs%5jSV(IJroMCQgq+jpl_A?GI zZUR38*pxy2I-Z>8!2TkY8b}a@v0S!CQT19#$g4*)dZtSnm5nJh*&}x?p(`TOJU;S2 z;t&H*3qn?;=uj>V2@sH&t!I(ixJ>;Yrgiz7W?#g4s4F5|eX&(Rj#!Ob)_PX2$Bi^r z4?7<<)b#CdP+p%SN;`+0~0LwfO!eDC_%wh`Wo0(jC1ZX3>>wC8|2B5eYCPJ81S%!}wE!4oefr zTxb2Gv+!K`*xbt&W4DO}F_Zw7ysQTVVu+fZQVE$I!&gf2aF<#}R;$S38lo2p(06yurJl2OL&w-rD+YP@v`gMjy| zb?)l_eYwNpwXU$zTI2Y^S#={Kd|f3OBNKr1~Kz1l55g7pa1~#TVX)} literal 0 HcmV?d00001 diff --git a/index.xml b/index.xml index 93e9d0701..5f21ea931 100644 --- a/index.xml +++ b/index.xml @@ -1,4 +1,4 @@ -Yanir Seroussi | Data & AI for Startup Impacthttps://yanirseroussi.com/Recent content on Yanir Seroussi | Data & AI for Startup ImpactHugo -- gohugo.ioen-auText and figures licensed under [CC BY-NC-ND 4.0](https://creativecommons.org/licenses/by-nc-nd/4.0/) by [Yanir Seroussi](https://yanirseroussi.com/about/), except where noted otherwiseMon, 03 Jun 2024 02:45:00 +0000How to avoid startups with poor development processeshttps://yanirseroussi.com/2024/06/03/how-to-avoid-startups-with-poor-development-processes/Mon, 03 Jun 2024 02:45:00 +0000https://yanirseroussi.com/2024/06/03/how-to-avoid-startups-with-poor-development-processes/Questions that prospective data specialists and engineers should ask about development processes before accepting a startup role.Plumbing, Decisions, and Automation: De-hyping Data & AIhttps://yanirseroussi.com/2024/05/27/plumbing-decisions-and-automation-de-hyping-data-and-ai/Mon, 27 May 2024 02:00:00 +0000https://yanirseroussi.com/2024/05/27/plumbing-decisions-and-automation-de-hyping-data-and-ai/Three essential questions to understand where an organisation stands when it comes to Data &amp; AI (with zero hype).Adapting to the economy of algorithmshttps://yanirseroussi.com/til/2024/05/25/adapting-to-the-economy-of-algorithms/Sat, 25 May 2024 00:00:00 +0000https://yanirseroussi.com/til/2024/05/25/adapting-to-the-economy-of-algorithms/Overview of the book The Economy of Algorithms by Marek Kowalkiewicz.Question startup culture before accepting a data-to-AI rolehttps://yanirseroussi.com/2024/05/20/question-startup-culture-before-accepting-a-data-to-ai-role/Mon, 20 May 2024 02:25:00 +0000https://yanirseroussi.com/2024/05/20/question-startup-culture-before-accepting-a-data-to-ai-role/Eight questions that prospective data-to-AI employees should ask about a startup&rsquo;s work and data culture.Probing the People aspects of an early-stage startuphttps://yanirseroussi.com/2024/05/13/probing-the-people-aspects-of-an-early-stage-startup/Mon, 13 May 2024 02:00:00 +0000https://yanirseroussi.com/2024/05/13/probing-the-people-aspects-of-an-early-stage-startup/Ten questions that prospective employees should ask about a startup&rsquo;s team, especially for data-centric roles.Business questions to ask before taking a startup data rolehttps://yanirseroussi.com/2024/05/06/business-questions-to-ask-before-taking-a-startup-data-role/Mon, 06 May 2024 04:30:00 +0000https://yanirseroussi.com/2024/05/06/business-questions-to-ask-before-taking-a-startup-data-role/Fourteen questions that prospective employees should ask about a startup&rsquo;s business model and product, especially for data-focused roles.Mentorship and the art of actionable advicehttps://yanirseroussi.com/2024/04/29/mentorship-and-the-art-of-actionable-advice/Mon, 29 Apr 2024 06:30:00 +0000https://yanirseroussi.com/2024/04/29/mentorship-and-the-art-of-actionable-advice/Reflections on what it takes to package expertise and deliver timely, actionable advice outside the context of employee relationships.Assessing a startup's data-to-AI healthhttps://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/Mon, 22 Apr 2024 06:00:00 +0000https://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/Reviewing the areas that should be assessed to determine a startup&rsquo;s opportunities and challenges on the data/AI/ML front.AI does not obviate the need for testing and observabilityhttps://yanirseroussi.com/2024/04/15/ai-does-not-obviate-the-need-for-testing-and-observability/Mon, 15 Apr 2024 05:00:00 +0000https://yanirseroussi.com/2024/04/15/ai-does-not-obviate-the-need-for-testing-and-observability/It&rsquo;s easy to prototype with AI, but production-grade AI apps require even more thorough testing and observability than traditional software.LinkedIn is a teachable skillhttps://yanirseroussi.com/til/2024/04/11/linkedin-is-a-teachable-skill/Thu, 11 Apr 2024 01:45:25 +0000https://yanirseroussi.com/til/2024/04/11/linkedin-is-a-teachable-skill/An high-level overview of things I learned from Justin Welsh&rsquo;s LinkedIn Operating System course.My experience as a Data Tech Lead with Work on Climatehttps://yanirseroussi.com/2024/04/08/my-experience-as-a-data-tech-lead-with-work-on-climate/Mon, 08 Apr 2024 02:00:00 +0000https://yanirseroussi.com/2024/04/08/my-experience-as-a-data-tech-lead-with-work-on-climate/The story of how I joined Work on Climate as a volunteer and became its data tech lead, with lessons applied to consulting &amp; fractional work.The data engineering lifecycle is not going anywherehttps://yanirseroussi.com/til/2024/04/05/the-data-engineering-lifecycle-is-not-going-anywhere/Fri, 05 Apr 2024 01:00:00 +0000https://yanirseroussi.com/til/2024/04/05/the-data-engineering-lifecycle-is-not-going-anywhere/My key takeaways from reading Fundamentals of Data Engineering by Joe Reis and Matt Housley.Artificial intelligence, automation, and the art of counting fishhttps://yanirseroussi.com/2024/04/01/artificial-intelligence-automation-and-the-art-of-counting-fish/Mon, 01 Apr 2024 06:00:00 +0000https://yanirseroussi.com/2024/04/01/artificial-intelligence-automation-and-the-art-of-counting-fish/Discussing the use of AI to automate underwater marine surveys as an example of the uneven distribution of technological advancement.Atomic Habits is full of actionable advicehttps://yanirseroussi.com/til/2024/03/12/atomic-habits-is-full-of-actionable-advice/Tue, 12 Mar 2024 06:19:31 +0000https://yanirseroussi.com/til/2024/03/12/atomic-habits-is-full-of-actionable-advice/I put the book to use after the first listen, and will definitely revisit it in the future to form better habits.Questions to consider when using AI for PDF data extractionhttps://yanirseroussi.com/2024/03/11/questions-to-consider-when-using-ai-for-pdf-data-extraction/Mon, 11 Mar 2024 00:00:00 +0000https://yanirseroussi.com/2024/03/11/questions-to-consider-when-using-ai-for-pdf-data-extraction/Discussing considerations that arise when attempting to automate the extraction of structured data from PDFs and similar documents.Two types of startup data problemshttps://yanirseroussi.com/2024/03/04/two-types-of-startup-data-problems/Mon, 04 Mar 2024 02:00:00 +0000https://yanirseroussi.com/2024/03/04/two-types-of-startup-data-problems/Classifying startups as ML-centric or non-ML is a helpful exercise to uncover the data challenges they&rsquo;re likely to face.Avoiding AI complexity: First, write no codehttps://yanirseroussi.com/2024/02/26/avoiding-ai-complexity-first-write-no-code/Mon, 26 Feb 2024 01:45:00 +0000https://yanirseroussi.com/2024/02/26/avoiding-ai-complexity-first-write-no-code/Two stories of getting AI functionality to production, which demonstrate the risks inherent in custom development versus starting with a no-code approach.Building your startup's minimum viable data stackhttps://yanirseroussi.com/2024/02/19/building-your-startups-minimum-viable-data-stack/Mon, 19 Feb 2024 00:00:00 +0000https://yanirseroussi.com/2024/02/19/building-your-startups-minimum-viable-data-stack/First post in a series on building a minimum viable data stack for startups, introducing key definitions, components, and considerations.The three Cs of indie consulting: Confidence, Cash, and Connectionshttps://yanirseroussi.com/til/2024/02/17/the-three-cs-of-indie-consulting-confidence-cash-and-connections/Sat, 17 Feb 2024 02:00:00 +0000https://yanirseroussi.com/til/2024/02/17/the-three-cs-of-indie-consulting-confidence-cash-and-connections/Jonathan Stark makes a compelling argument why you should have the three Cs before quitting your job to go solo consulting.Nudging ChatGPT to invent books you have no time to readhttps://yanirseroussi.com/2024/02/12/nudging-chatgpt-to-invent-books-you-have-no-time-to-read/Mon, 12 Feb 2024 05:00:00 +0000https://yanirseroussi.com/2024/02/12/nudging-chatgpt-to-invent-books-you-have-no-time-to-read/Getting ChatGPT Plus to elaborate on possible book content and produce a PDF cheatsheet, with the goal of learning about its capabilities.Future software development may require fewer humanshttps://yanirseroussi.com/til/2024/02/06/future-software-development-may-require-fewer-humans/Tue, 06 Feb 2024 06:15:00 +0000https://yanirseroussi.com/til/2024/02/06/future-software-development-may-require-fewer-humans/Reflecting on an interview with Jason Warner, CEO of poolside.Substance over titles: Your first data hire may be a data scientisthttps://yanirseroussi.com/2024/02/05/substance-over-titles-your-first-data-hire-may-be-a-data-scientist/Mon, 05 Feb 2024 02:45:00 +0000https://yanirseroussi.com/2024/02/05/substance-over-titles-your-first-data-hire-may-be-a-data-scientist/Advice for hiring a startup&rsquo;s first data person: match skills to business needs, consider contractors, and get help from data people.New decade, new tagline: Data & AI for Impacthttps://yanirseroussi.com/2024/01/19/new-decade-new-tagline-data-and-ai-for-impact/Fri, 19 Jan 2024 00:00:00 +0000https://yanirseroussi.com/2024/01/19/new-decade-new-tagline-data-and-ai-for-impact/Shifting focus to &lsquo;Data &amp; AI for Impact&rsquo;, with more startup-related content, increased posting frequency, and deeper audience engagement.Psychographic specialisations may work for discipline generalistshttps://yanirseroussi.com/til/2024/01/09/psychographic-specialisations-may-work-for-discipline-generalists/Tue, 09 Jan 2024 03:00:00 +0000https://yanirseroussi.com/til/2024/01/09/psychographic-specialisations-may-work-for-discipline-generalists/When focusing on a market segment defined by personal beliefs, it&rsquo;s often fine to position yourself as a generalist in your craft.The power of parasocial relationshipshttps://yanirseroussi.com/til/2024/01/08/the-power-of-parasocial-relationships/Mon, 08 Jan 2024 06:00:00 +0000https://yanirseroussi.com/til/2024/01/08/the-power-of-parasocial-relationships/Repeated exposure to media personas creates relationships that help justify premium fees.Positioning is a common problem for data scientistshttps://yanirseroussi.com/til/2023/12/18/positioning-is-a-common-problem-for-data-scientists/Mon, 18 Dec 2023 00:30:00 +0000https://yanirseroussi.com/til/2023/12/18/positioning-is-a-common-problem-for-data-scientists/With the commodification of data scientists, the problem of positioning has become more common: My takeaways from Genevieve Hayes interviewing Jonathan Stark.Transfer learning applies to energy market biddinghttps://yanirseroussi.com/til/2023/12/14/transfer-learning-applies-to-energy-market-bidding/Thu, 14 Dec 2023 00:15:00 +0000https://yanirseroussi.com/til/2023/12/14/transfer-learning-applies-to-energy-market-bidding/An interesting approach to bidding of energy storage assets, showing that training on New York data is transferable to Queensland.Supporting volunteer monitoring of marine biodiversity with modern web and data toolshttps://yanirseroussi.com/2023/11/29/supporting-volunteer-monitoring-of-marine-biodiversity-with-modern-web-and-data-tools/Wed, 29 Nov 2023 02:00:00 +0000https://yanirseroussi.com/2023/11/29/supporting-volunteer-monitoring-of-marine-biodiversity-with-modern-web-and-data-tools/Summarising the work Uri Seroussi and I did to improve Reef Life Survey&rsquo;s Reef Species of the World app.Our Blue Machine is changing, but we are not helplesshttps://yanirseroussi.com/til/2023/11/28/our-blue-machine-is-changing-but-we-are-not-helpless/Tue, 28 Nov 2023 06:40:00 +0000https://yanirseroussi.com/til/2023/11/28/our-blue-machine-is-changing-but-we-are-not-helpless/One of my many highlights from Helen Czerski&rsquo;s Blue Machine.You don't need a proprietary API for static mapshttps://yanirseroussi.com/til/2023/11/21/you-dont-need-a-proprietary-api-for-static-maps/Tue, 21 Nov 2023 06:00:00 +0000https://yanirseroussi.com/til/2023/11/21/you-dont-need-a-proprietary-api-for-static-maps/For many use cases, libraries like cartopy are better than the likes of Mapbox and Google Maps.Lessons from reluctant data engineeringhttps://yanirseroussi.com/2023/10/25/lessons-from-reluctant-data-engineering/Wed, 25 Oct 2023 04:45:00 +0000https://yanirseroussi.com/2023/10/25/lessons-from-reluctant-data-engineering/Video and summary of a talk I gave at DataEngBytes Brisbane on what I learned from doing data engineering as part of every data science role I had.Artificial intelligence was a marketing term all along – just call it automationhttps://yanirseroussi.com/til/2023/10/06/artificial-intelligence-was-a-marketing-term-all-along-just-call-it-automation/Fri, 06 Oct 2023 05:00:00 +0000https://yanirseroussi.com/til/2023/10/06/artificial-intelligence-was-a-marketing-term-all-along-just-call-it-automation/Replacing &lsquo;artificial intelligence&rsquo; with &lsquo;automation&rsquo; is a useful trick for cutting through the hype.The lines between solo consulting and product building are blurryhttps://yanirseroussi.com/til/2023/09/25/the-lines-between-solo-consulting-and-product-building-are-blurry/Mon, 25 Sep 2023 00:00:00 +0000https://yanirseroussi.com/til/2023/09/25/the-lines-between-solo-consulting-and-product-building-are-blurry/It turns out that problems like finding a niche and defining the ideal clients are key to any solo business.Google's Rules of Machine Learning still apply in the age of large language modelshttps://yanirseroussi.com/til/2023/09/21/googles-rules-of-machine-learning-still-apply-in-the-age-of-large-language-models/Thu, 21 Sep 2023 21:30:00 +0000https://yanirseroussi.com/til/2023/09/21/googles-rules-of-machine-learning-still-apply-in-the-age-of-large-language-models/Despite the excitement around large language models, building with machine learning remains an engineering problem with established best practices.My rediscovery of quiet writing on the open webhttps://yanirseroussi.com/2023/08/28/my-rediscovery-of-quiet-writing-on-the-open-web/Mon, 28 Aug 2023 05:30:00 +0000https://yanirseroussi.com/2023/08/28/my-rediscovery-of-quiet-writing-on-the-open-web/Reflections on publishing on this website: Writing publicly to share thoughts and documentation beats chasing views and likes.The Minimalist Entrepreneur is too prescriptive for mehttps://yanirseroussi.com/til/2023/08/21/the-minimalist-entrepreneur-is-too-prescriptive-for-me/Mon, 21 Aug 2023 03:15:00 +0000https://yanirseroussi.com/til/2023/08/21/the-minimalist-entrepreneur-is-too-prescriptive-for-me/While I found the story of Gumroad interesting, The Minimalist Entrepreneur seems to over-generalise from the founder&rsquo;s experience.Revisiting Start Small, Stay Small in 2023 (Chapter 2)https://yanirseroussi.com/til/2023/08/17/revisiting-start-small-stay-small-in-2023-chapter-2/Thu, 17 Aug 2023 07:45:00 +0000https://yanirseroussi.com/til/2023/08/17/revisiting-start-small-stay-small-in-2023-chapter-2/A summary of the second chapter of Rob Walling&rsquo;s Start Small, Stay Small, along with my thoughts &amp; reflections.Revisiting Start Small, Stay Small in 2023 (Chapter 1)https://yanirseroussi.com/til/2023/08/16/revisiting-start-small-stay-small-in-2023-chapter-1/Wed, 16 Aug 2023 05:45:00 +0000https://yanirseroussi.com/til/2023/08/16/revisiting-start-small-stay-small-in-2023-chapter-1/A summary of the first chapter of Rob Walling&rsquo;s Start Small, Stay Small, along with my thoughts &amp; reflections.Email notifications on public GitHub commitshttps://yanirseroussi.com/til/2023/08/14/email-notifications-on-public-github-commits/Mon, 14 Aug 2023 05:15:00 +0000https://yanirseroussi.com/til/2023/08/14/email-notifications-on-public-github-commits/GitHub publishes an Atom feed, which means you can use any RSS reader to follow commits.The rule of thirds can probably be ignoredhttps://yanirseroussi.com/til/2023/08/11/the-rule-of-thirds-can-probably-be-ignored/Fri, 11 Aug 2023 03:15:00 +0000https://yanirseroussi.com/til/2023/08/11/the-rule-of-thirds-can-probably-be-ignored/Turns out that the rule of thirds for composing visuals may not be that important.Using YubiKey for SSH accesshttps://yanirseroussi.com/til/2023/07/23/using-yubikey-for-ssh-access/Sun, 23 Jul 2023 00:07:15 +0000https://yanirseroussi.com/til/2023/07/23/using-yubikey-for-ssh-access/Some pointers for setting up SSH access with YubiKey on Ubuntu 22.04.Making a TIL section with Hugo and PaperModhttps://yanirseroussi.com/til/2023/07/17/making-a-til-section-with-hugo-and-papermod/Mon, 17 Jul 2023 00:06:15 +0000https://yanirseroussi.com/til/2023/07/17/making-a-til-section-with-hugo-and-papermod/How I added a Today I Learned section to my Hugo site with the PaperMod theme.You can't save timehttps://yanirseroussi.com/til/2023/07/11/you-cant-save-time/Tue, 11 Jul 2023 00:00:00 +0000https://yanirseroussi.com/til/2023/07/11/you-cant-save-time/Time can be spent doing different activities, but it can&rsquo;t be stored and saved for later.Was data science a failure mode of software engineering?https://yanirseroussi.com/2023/06/30/was-data-science-a-failure-mode-of-software-engineering/Fri, 30 Jun 2023 00:06:30 +0000https://yanirseroussi.com/2023/06/30/was-data-science-a-failure-mode-of-software-engineering/Yes, data science projects have suffered from classic software engineering mistakes, but the field is maturing with the rise of new engineering roles.How hackable are automated coding assessments?https://yanirseroussi.com/2023/05/26/how-hackable-are-automated-coding-assessments/Fri, 26 May 2023 00:03:00 +0000https://yanirseroussi.com/2023/05/26/how-hackable-are-automated-coding-assessments/Exploring the hackability of speed-based coding tests, using CodeSignal&rsquo;s Industry Coding Framework as a case study.Remaining relevant as a small language modelhttps://yanirseroussi.com/2023/04/21/remaining-relevant-as-a-small-language-model/Fri, 21 Apr 2023 00:06:30 +0000https://yanirseroussi.com/2023/04/21/remaining-relevant-as-a-small-language-model/Bing Chat recently quipped that humans are small language models. Here are some of my thoughts on how we small language models can remain relevant (for now).ChatGPT is transformative AIhttps://yanirseroussi.com/2022/12/11/chatgpt-is-transformative-ai/Sun, 11 Dec 2022 00:00:00 +0000https://yanirseroussi.com/2022/12/11/chatgpt-is-transformative-ai/My perspective after a week of using ChatGPT: This is a step change in finding distilled information, and it&rsquo;s only the beginning.Causal Machine Learning is off to a good start, despite some issueshttps://yanirseroussi.com/2022/09/12/causal-machine-learning-book-draft-review/Mon, 12 Sep 2022 02:45:00 +0000https://yanirseroussi.com/2022/09/12/causal-machine-learning-book-draft-review/Reviewing the first three chapters of the book Causal Machine Learning by Robert Osazuwa Ness.The mission matters: Moving to climate tech as a data scientisthttps://yanirseroussi.com/2022/06/06/the-mission-matters-moving-to-climate-tech-as-a-data-scientist/Mon, 06 Jun 2022 00:00:00 +0000https://yanirseroussi.com/2022/06/06/the-mission-matters-moving-to-climate-tech-as-a-data-scientist/Discussing my recent career move into climate tech as a way of doing more to help mitigate dangerous climate change.Building useful machine learning tools keeps getting easier: A fish ID case studyhttps://yanirseroussi.com/2022/03/20/building-useful-machine-learning-tools-keeps-getting-easier-a-fish-id-case-study/Sun, 20 Mar 2022 04:30:00 +0000https://yanirseroussi.com/2022/03/20/building-useful-machine-learning-tools-keeps-getting-easier-a-fish-id-case-study/Lessons learned building a fish ID web app with fast.ai and Streamlit, in an attempt to reduce my fear of missing out on the latest deep learning developments.Analysis strategies in online A/B experiments: Intention-to-treat, per-protocol, and other lessons from clinical trialshttps://yanirseroussi.com/2022/01/14/analysis-strategies-in-online-a-b-experiments/Fri, 14 Jan 2022 00:05:40 +0000https://yanirseroussi.com/2022/01/14/analysis-strategies-in-online-a-b-experiments/Epidemiologists analyse clinical trials to estimate the intention-to-treat and per-protocol effects. This post applies their strategies to online experiments.Use your human brain to avoid artificial intelligence disastershttps://yanirseroussi.com/2021/11/22/use-your-human-brain-to-avoid-artificial-intelligence-disasters/Mon, 22 Nov 2021 03:45:00 +0000https://yanirseroussi.com/2021/11/22/use-your-human-brain-to-avoid-artificial-intelligence-disasters/Overview of a talk I gave at a deep learning course, focusing on AI ethics as the need for humans to think on the context and consequences of applying AI.Migrating from WordPress.com to Hugo on GitHub + Cloudflarehttps://yanirseroussi.com/2021/11/10/migrating-from-wordpress-com-to-hugo-on-github-cloudflare/Wed, 10 Nov 2021 06:30:00 +0000https://yanirseroussi.com/2021/11/10/migrating-from-wordpress-com-to-hugo-on-github-cloudflare/My reasons for switching from WordPress.com to Hugo on GitHub + Cloudflare, along with a summary of the solution components and migration process.My work with Automattichttps://yanirseroussi.com/2021/10/07/my-work-with-automattic/Thu, 07 Oct 2021 00:00:00 +0000https://yanirseroussi.com/2021/10/07/my-work-with-automattic/Back-dated meta-post that gathers my posts on Automattic blogs into a summary of the work I&rsquo;ve done with the company.Some highlights from 2020https://yanirseroussi.com/2021/04/05/some-highlights-from-2020/Mon, 05 Apr 2021 06:41:48 +0000https://yanirseroussi.com/2021/04/05/some-highlights-from-2020/Sharing remote teamwork insights, my climate &amp; sustainability activism, Reef Life Survey publications, and progress on Automattic&rsquo;s Experimentation Platform.Many is not enough: Counting simulations to bootstrap the right wayhttps://yanirseroussi.com/2020/08/24/many-is-not-enough-counting-simulations-to-bootstrap-the-right-way/Mon, 24 Aug 2020 01:35:17 +0000https://yanirseroussi.com/2020/08/24/many-is-not-enough-counting-simulations-to-bootstrap-the-right-way/Going deeper into correct testing of different methods for bootstrap estimation of confidence intervals.Software commodities are eating interesting data science workhttps://yanirseroussi.com/2020/01/11/software-commodities-are-eating-interesting-data-science-work/Sat, 11 Jan 2020 09:22:35 +0000https://yanirseroussi.com/2020/01/11/software-commodities-are-eating-interesting-data-science-work/Being a data scientist can sometimes feel like a race against software commodities that replace interesting work. What can one do to remain relevant?A day in the life of a remote data scientisthttps://yanirseroussi.com/2019/12/12/a-day-in-the-life-of-a-remote-data-scientist/Wed, 11 Dec 2019 22:06:19 +0000https://yanirseroussi.com/2019/12/12/a-day-in-the-life-of-a-remote-data-scientist/Video of a talk I gave on remote data science work at the Data Science Sydney meetup.Bootstrapping the right way?https://yanirseroussi.com/2019/10/06/bootstrapping-the-right-way/Sun, 06 Oct 2019 06:48:07 +0000https://yanirseroussi.com/2019/10/06/bootstrapping-the-right-way/Video and summary of a talk I gave at YOW! Data on bootstrap estimation of confidence intervals.Hackers beware: Bootstrap sampling may be harmfulhttps://yanirseroussi.com/2019/01/08/hackers-beware-bootstrap-sampling-may-be-harmful/Mon, 07 Jan 2019 21:07:56 +0000https://yanirseroussi.com/2019/01/08/hackers-beware-bootstrap-sampling-may-be-harmful/Bootstrap sampling has been promoted as an easy way of modelling uncertainty to hackers without much statistical knowledge. But things aren&rsquo;t that simple.The most practical causal inference book I’ve read (is still a draft)https://yanirseroussi.com/2018/12/24/the-most-practical-causal-inference-book-ive-read-is-still-a-draft/Mon, 24 Dec 2018 02:37:50 +0000https://yanirseroussi.com/2018/12/24/the-most-practical-causal-inference-book-ive-read-is-still-a-draft/Causal Inference by Miguel Hernán and Jamie Robins is a must-read for anyone interested in the area.Reflections on remote data science workhttps://yanirseroussi.com/2018/11/03/reflections-on-remote-data-science-work/Sat, 03 Nov 2018 06:33:13 +0000https://yanirseroussi.com/2018/11/03/reflections-on-remote-data-science-work/Discussing the pluses and minuses of remote work eighteen months after joining Automattic as a data scientist.Defining data science in 2018https://yanirseroussi.com/2018/07/22/defining-data-science-in-2018/Sun, 22 Jul 2018 08:27:43 +0000https://yanirseroussi.com/2018/07/22/defining-data-science-in-2018/Updating my definition of data science to match changes in the field. It is now broader than before, but its ultimate goal is still to support decisions.Advice for aspiring data scientists and other FAQshttps://yanirseroussi.com/2017/10/15/advice-for-aspiring-data-scientists-and-other-faqs/Sun, 15 Oct 2017 09:15:25 +0000https://yanirseroussi.com/2017/10/15/advice-for-aspiring-data-scientists-and-other-faqs/Frequently asked questions by visitors to this site, especially around entering the data science field.State of Bandcamp Recommender, Late 2017https://yanirseroussi.com/2017/09/02/state-of-bandcamp-recommender/Sat, 02 Sep 2017 10:19:02 +0000https://yanirseroussi.com/2017/09/02/state-of-bandcamp-recommender/Call for BCRecommender maintainers followed by a decision to shut it down, as I don&rsquo;t have enough time and Bandcamp now offers recommendations.My 10-step path to becoming a remote data scientist with Automattichttps://yanirseroussi.com/2017/07/29/my-10-step-path-to-becoming-a-remote-data-scientist-with-automattic/Sat, 29 Jul 2017 05:39:26 +0000https://yanirseroussi.com/2017/07/29/my-10-step-path-to-becoming-a-remote-data-scientist-with-automattic/I wanted a well-paid data science-y remote job with an established company that offers a good life balance and makes products I care about. I got it eventually.Exploring and visualising Reef Life Survey datahttps://yanirseroussi.com/2017/06/03/exploring-and-visualising-reef-life-survey-data/Sat, 03 Jun 2017 00:49:05 +0000https://yanirseroussi.com/2017/06/03/exploring-and-visualising-reef-life-survey-data/Web tools I built to visualise Reef Life Survey data and assist citizen scientists in underwater visual census work.Customer lifetime value and the proliferation of misinformation on the internethttps://yanirseroussi.com/2017/01/08/customer-lifetime-value-and-the-proliferation-of-misinformation-on-the-internet/Sun, 08 Jan 2017 20:02:30 +0000https://yanirseroussi.com/2017/01/08/customer-lifetime-value-and-the-proliferation-of-misinformation-on-the-internet/There&rsquo;s a lot of misleading content on the estimation of customer lifetime value. Here&rsquo;s what I learned about doing it well.Ask Why! Finding motives, causes, and purpose in data sciencehttps://yanirseroussi.com/2016/09/19/ask-why-finding-motives-causes-and-purpose-in-data-science/Mon, 19 Sep 2016 21:28:44 +0000https://yanirseroussi.com/2016/09/19/ask-why-finding-motives-causes-and-purpose-in-data-science/Video and summary of a talk I gave at the Data Science Sydney meetup, about going beyond the what &amp; how of predictive modelling.If you don’t pay attention, data can drive you off a cliffhttps://yanirseroussi.com/2016/08/21/seven-ways-to-be-data-driven-off-a-cliff/Sun, 21 Aug 2016 21:34:17 +0000https://yanirseroussi.com/2016/08/21/seven-ways-to-be-data-driven-off-a-cliff/Seven common mistakes to avoid when working with data, such as ignoring uncertainty and confusing observed and unobserved quantities.Is Data Scientist a useless job title?https://yanirseroussi.com/2016/08/04/is-data-scientist-a-useless-job-title/Thu, 04 Aug 2016 22:26:03 +0000https://yanirseroussi.com/2016/08/04/is-data-scientist-a-useless-job-title/It seems like anyone who touches data can call themselves a data scientist, which makes the title useless. The work they do can still be useful, though.Making Bayesian A/B testing more accessiblehttps://yanirseroussi.com/2016/06/19/making-bayesian-ab-testing-more-accessible/Sun, 19 Jun 2016 10:32:15 +0000https://yanirseroussi.com/2016/06/19/making-bayesian-ab-testing-more-accessible/A web tool I built to interpret A/B test results in a Bayesian way, including prior specification, visualisations, and decision rules.Diving deeper into causality: Pearl, Kleinberg, Hill, and untested assumptionshttps://yanirseroussi.com/2016/05/15/diving-deeper-into-causality-pearl-kleinberg-hill-and-untested-assumptions/Sat, 14 May 2016 19:57:03 +0000https://yanirseroussi.com/2016/05/15/diving-deeper-into-causality-pearl-kleinberg-hill-and-untested-assumptions/Discussing the need for untested assumptions and temporality in causal inference. Mostly based on Samantha Kleinberg&rsquo;s Causality, Probability, and Time.The rise of greedy robotshttps://yanirseroussi.com/2016/03/20/the-rise-of-greedy-robots/Sun, 20 Mar 2016 20:33:43 +0000https://yanirseroussi.com/2016/03/20/the-rise-of-greedy-robots/Is artificial/machine intelligence a future threat? I argue that it&rsquo;s already here, with greedy robots already dominating our lives.Why you should stop worrying about deep learning and deepen your understanding of causality insteadhttps://yanirseroussi.com/2016/02/14/why-you-should-stop-worrying-about-deep-learning-and-deepen-your-understanding-of-causality-instead/Sun, 14 Feb 2016 11:04:11 +0000https://yanirseroussi.com/2016/02/14/why-you-should-stop-worrying-about-deep-learning-and-deepen-your-understanding-of-causality-instead/Causality is often overlooked but is of much higher relevance to most data scientists than deep learning.The joys of offline data collectionhttps://yanirseroussi.com/2016/01/24/the-joys-of-offline-data-collection/Sun, 24 Jan 2016 00:32:25 +0000https://yanirseroussi.com/2016/01/24/the-joys-of-offline-data-collection/Insights on data collection and machine learning from spending a month sailing, diving, and counting fish with Reef Life Survey.This holiday season, give me real insightshttps://yanirseroussi.com/2015/12/08/this-holiday-season-give-me-real-insights/Tue, 08 Dec 2015 06:57:25 +0000https://yanirseroussi.com/2015/12/08/this-holiday-season-give-me-real-insights/Some companies present raw data or information as &ldquo;insights&rdquo;. This post surveys some examples, and discusses how they can be turned into real insights.The hardest parts of data sciencehttps://yanirseroussi.com/2015/11/23/the-hardest-parts-of-data-science/Mon, 23 Nov 2015 04:14:21 +0000https://yanirseroussi.com/2015/11/23/the-hardest-parts-of-data-science/Defining feasible problems and coming up with reasonable ways of measuring solutions is harder than building accurate models or obtaining clean data.Migrating a simple web application from MongoDB to Elasticsearchhttps://yanirseroussi.com/2015/11/04/migrating-a-simple-web-application-from-mongodb-to-elasticsearch/Wed, 04 Nov 2015 03:53:18 +0000https://yanirseroussi.com/2015/11/04/migrating-a-simple-web-application-from-mongodb-to-elasticsearch/Migrating BCRecommender from MongoDB to Elasticsearch made it possible to offer a richer search experience to users at a similar cost, among other benefits.Miscommunicating science: Simplistic models, nutritionism, and the art of storytellinghttps://yanirseroussi.com/2015/10/19/nutritionism-and-the-need-for-complex-models-to-explain-complex-phenomena/Mon, 19 Oct 2015 00:02:32 +0000https://yanirseroussi.com/2015/10/19/nutritionism-and-the-need-for-complex-models-to-explain-complex-phenomena/Nutritionism is a special case of misinterpretation and miscommunication of scientific results – something many data scientists encounter in their work.The wonderful world of recommender systemshttps://yanirseroussi.com/2015/10/02/the-wonderful-world-of-recommender-systems/Fri, 02 Oct 2015 05:25:57 +0000https://yanirseroussi.com/2015/10/02/the-wonderful-world-of-recommender-systems/Giving an overview of the field and common paradigms, and debunking five common myths about recommender systems.You don’t need a data scientist (yet)https://yanirseroussi.com/2015/08/24/you-dont-need-a-data-scientist-yet/Mon, 24 Aug 2015 08:25:30 +0000https://yanirseroussi.com/2015/08/24/you-dont-need-a-data-scientist-yet/Hiring data scientists prematurely is wasteful and frustrating. Here are some questions to ask before you hire your first data scientist.Goodbye, Parse.comhttps://yanirseroussi.com/2015/07/31/goodbye-parse-com/Fri, 31 Jul 2015 03:29:50 +0000https://yanirseroussi.com/2015/07/31/goodbye-parse-com/Migrating my web apps away from Parse.com due to reliability issues. Self-hosting is a better solution.Learning about deep learning through album cover classificationhttps://yanirseroussi.com/2015/07/06/learning-about-deep-learning-through-album-cover-classification/Mon, 06 Jul 2015 22:21:42 +0000https://yanirseroussi.com/2015/07/06/learning-about-deep-learning-through-album-cover-classification/Progress on my album cover classification project, highlighting lessons that would be useful to others who are getting started with deep learning.Deep learning resourceshttps://yanirseroussi.com/deep-learning-resources/Mon, 06 Jul 2015 00:38:44 +0000https://yanirseroussi.com/deep-learning-resources/This page summarises the deep learning resources I&rsquo;ve consulted in my album cover classification project. +Yanir Seroussi | Data & AI for Startup Impacthttps://yanirseroussi.com/Recent content on Yanir Seroussi | Data & AI for Startup ImpactHugo -- gohugo.ioen-auText and figures licensed under [CC BY-NC-ND 4.0](https://creativecommons.org/licenses/by-nc-nd/4.0/) by [Yanir Seroussi](https://yanirseroussi.com/about/), except where noted otherwiseMon, 10 Jun 2024 04:00:00 +0000Startup data health starts with healthy event trackinghttps://yanirseroussi.com/2024/06/10/startup-data-health-starts-with-healthy-event-tracking/Mon, 10 Jun 2024 04:00:00 +0000https://yanirseroussi.com/2024/06/10/startup-data-health-starts-with-healthy-event-tracking/Expanding on the startup health check question of tracking Kukuyeva&rsquo;s five business aspects as wide events.How to avoid startups with poor development processeshttps://yanirseroussi.com/2024/06/03/how-to-avoid-startups-with-poor-development-processes/Mon, 03 Jun 2024 02:45:00 +0000https://yanirseroussi.com/2024/06/03/how-to-avoid-startups-with-poor-development-processes/Questions that prospective data specialists and engineers should ask about development processes before accepting a startup role.Plumbing, Decisions, and Automation: De-hyping Data & AIhttps://yanirseroussi.com/2024/05/27/plumbing-decisions-and-automation-de-hyping-data-and-ai/Mon, 27 May 2024 02:00:00 +0000https://yanirseroussi.com/2024/05/27/plumbing-decisions-and-automation-de-hyping-data-and-ai/Three essential questions to understand where an organisation stands when it comes to Data &amp; AI (with zero hype).Adapting to the economy of algorithmshttps://yanirseroussi.com/til/2024/05/25/adapting-to-the-economy-of-algorithms/Sat, 25 May 2024 00:00:00 +0000https://yanirseroussi.com/til/2024/05/25/adapting-to-the-economy-of-algorithms/Overview of the book The Economy of Algorithms by Marek Kowalkiewicz.Question startup culture before accepting a data-to-AI rolehttps://yanirseroussi.com/2024/05/20/question-startup-culture-before-accepting-a-data-to-ai-role/Mon, 20 May 2024 02:25:00 +0000https://yanirseroussi.com/2024/05/20/question-startup-culture-before-accepting-a-data-to-ai-role/Eight questions that prospective data-to-AI employees should ask about a startup&rsquo;s work and data culture.Probing the People aspects of an early-stage startuphttps://yanirseroussi.com/2024/05/13/probing-the-people-aspects-of-an-early-stage-startup/Mon, 13 May 2024 02:00:00 +0000https://yanirseroussi.com/2024/05/13/probing-the-people-aspects-of-an-early-stage-startup/Ten questions that prospective employees should ask about a startup&rsquo;s team, especially for data-centric roles.Business questions to ask before taking a startup data rolehttps://yanirseroussi.com/2024/05/06/business-questions-to-ask-before-taking-a-startup-data-role/Mon, 06 May 2024 04:30:00 +0000https://yanirseroussi.com/2024/05/06/business-questions-to-ask-before-taking-a-startup-data-role/Fourteen questions that prospective employees should ask about a startup&rsquo;s business model and product, especially for data-focused roles.Mentorship and the art of actionable advicehttps://yanirseroussi.com/2024/04/29/mentorship-and-the-art-of-actionable-advice/Mon, 29 Apr 2024 06:30:00 +0000https://yanirseroussi.com/2024/04/29/mentorship-and-the-art-of-actionable-advice/Reflections on what it takes to package expertise and deliver timely, actionable advice outside the context of employee relationships.Assessing a startup's data-to-AI healthhttps://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/Mon, 22 Apr 2024 06:00:00 +0000https://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/Reviewing the areas that should be assessed to determine a startup&rsquo;s opportunities and challenges on the data/AI/ML front.AI does not obviate the need for testing and observabilityhttps://yanirseroussi.com/2024/04/15/ai-does-not-obviate-the-need-for-testing-and-observability/Mon, 15 Apr 2024 05:00:00 +0000https://yanirseroussi.com/2024/04/15/ai-does-not-obviate-the-need-for-testing-and-observability/It&rsquo;s easy to prototype with AI, but production-grade AI apps require even more thorough testing and observability than traditional software.LinkedIn is a teachable skillhttps://yanirseroussi.com/til/2024/04/11/linkedin-is-a-teachable-skill/Thu, 11 Apr 2024 01:45:25 +0000https://yanirseroussi.com/til/2024/04/11/linkedin-is-a-teachable-skill/An high-level overview of things I learned from Justin Welsh&rsquo;s LinkedIn Operating System course.My experience as a Data Tech Lead with Work on Climatehttps://yanirseroussi.com/2024/04/08/my-experience-as-a-data-tech-lead-with-work-on-climate/Mon, 08 Apr 2024 02:00:00 +0000https://yanirseroussi.com/2024/04/08/my-experience-as-a-data-tech-lead-with-work-on-climate/The story of how I joined Work on Climate as a volunteer and became its data tech lead, with lessons applied to consulting &amp; fractional work.The data engineering lifecycle is not going anywherehttps://yanirseroussi.com/til/2024/04/05/the-data-engineering-lifecycle-is-not-going-anywhere/Fri, 05 Apr 2024 01:00:00 +0000https://yanirseroussi.com/til/2024/04/05/the-data-engineering-lifecycle-is-not-going-anywhere/My key takeaways from reading Fundamentals of Data Engineering by Joe Reis and Matt Housley.Artificial intelligence, automation, and the art of counting fishhttps://yanirseroussi.com/2024/04/01/artificial-intelligence-automation-and-the-art-of-counting-fish/Mon, 01 Apr 2024 06:00:00 +0000https://yanirseroussi.com/2024/04/01/artificial-intelligence-automation-and-the-art-of-counting-fish/Discussing the use of AI to automate underwater marine surveys as an example of the uneven distribution of technological advancement.Atomic Habits is full of actionable advicehttps://yanirseroussi.com/til/2024/03/12/atomic-habits-is-full-of-actionable-advice/Tue, 12 Mar 2024 06:19:31 +0000https://yanirseroussi.com/til/2024/03/12/atomic-habits-is-full-of-actionable-advice/I put the book to use after the first listen, and will definitely revisit it in the future to form better habits.Questions to consider when using AI for PDF data extractionhttps://yanirseroussi.com/2024/03/11/questions-to-consider-when-using-ai-for-pdf-data-extraction/Mon, 11 Mar 2024 00:00:00 +0000https://yanirseroussi.com/2024/03/11/questions-to-consider-when-using-ai-for-pdf-data-extraction/Discussing considerations that arise when attempting to automate the extraction of structured data from PDFs and similar documents.Two types of startup data problemshttps://yanirseroussi.com/2024/03/04/two-types-of-startup-data-problems/Mon, 04 Mar 2024 02:00:00 +0000https://yanirseroussi.com/2024/03/04/two-types-of-startup-data-problems/Classifying startups as ML-centric or non-ML is a helpful exercise to uncover the data challenges they&rsquo;re likely to face.Avoiding AI complexity: First, write no codehttps://yanirseroussi.com/2024/02/26/avoiding-ai-complexity-first-write-no-code/Mon, 26 Feb 2024 01:45:00 +0000https://yanirseroussi.com/2024/02/26/avoiding-ai-complexity-first-write-no-code/Two stories of getting AI functionality to production, which demonstrate the risks inherent in custom development versus starting with a no-code approach.Building your startup's minimum viable data stackhttps://yanirseroussi.com/2024/02/19/building-your-startups-minimum-viable-data-stack/Mon, 19 Feb 2024 00:00:00 +0000https://yanirseroussi.com/2024/02/19/building-your-startups-minimum-viable-data-stack/First post in a series on building a minimum viable data stack for startups, introducing key definitions, components, and considerations.The three Cs of indie consulting: Confidence, Cash, and Connectionshttps://yanirseroussi.com/til/2024/02/17/the-three-cs-of-indie-consulting-confidence-cash-and-connections/Sat, 17 Feb 2024 02:00:00 +0000https://yanirseroussi.com/til/2024/02/17/the-three-cs-of-indie-consulting-confidence-cash-and-connections/Jonathan Stark makes a compelling argument why you should have the three Cs before quitting your job to go solo consulting.Nudging ChatGPT to invent books you have no time to readhttps://yanirseroussi.com/2024/02/12/nudging-chatgpt-to-invent-books-you-have-no-time-to-read/Mon, 12 Feb 2024 05:00:00 +0000https://yanirseroussi.com/2024/02/12/nudging-chatgpt-to-invent-books-you-have-no-time-to-read/Getting ChatGPT Plus to elaborate on possible book content and produce a PDF cheatsheet, with the goal of learning about its capabilities.Future software development may require fewer humanshttps://yanirseroussi.com/til/2024/02/06/future-software-development-may-require-fewer-humans/Tue, 06 Feb 2024 06:15:00 +0000https://yanirseroussi.com/til/2024/02/06/future-software-development-may-require-fewer-humans/Reflecting on an interview with Jason Warner, CEO of poolside.Substance over titles: Your first data hire may be a data scientisthttps://yanirseroussi.com/2024/02/05/substance-over-titles-your-first-data-hire-may-be-a-data-scientist/Mon, 05 Feb 2024 02:45:00 +0000https://yanirseroussi.com/2024/02/05/substance-over-titles-your-first-data-hire-may-be-a-data-scientist/Advice for hiring a startup&rsquo;s first data person: match skills to business needs, consider contractors, and get help from data people.New decade, new tagline: Data & AI for Impacthttps://yanirseroussi.com/2024/01/19/new-decade-new-tagline-data-and-ai-for-impact/Fri, 19 Jan 2024 00:00:00 +0000https://yanirseroussi.com/2024/01/19/new-decade-new-tagline-data-and-ai-for-impact/Shifting focus to &lsquo;Data &amp; AI for Impact&rsquo;, with more startup-related content, increased posting frequency, and deeper audience engagement.Psychographic specialisations may work for discipline generalistshttps://yanirseroussi.com/til/2024/01/09/psychographic-specialisations-may-work-for-discipline-generalists/Tue, 09 Jan 2024 03:00:00 +0000https://yanirseroussi.com/til/2024/01/09/psychographic-specialisations-may-work-for-discipline-generalists/When focusing on a market segment defined by personal beliefs, it&rsquo;s often fine to position yourself as a generalist in your craft.The power of parasocial relationshipshttps://yanirseroussi.com/til/2024/01/08/the-power-of-parasocial-relationships/Mon, 08 Jan 2024 06:00:00 +0000https://yanirseroussi.com/til/2024/01/08/the-power-of-parasocial-relationships/Repeated exposure to media personas creates relationships that help justify premium fees.Positioning is a common problem for data scientistshttps://yanirseroussi.com/til/2023/12/18/positioning-is-a-common-problem-for-data-scientists/Mon, 18 Dec 2023 00:30:00 +0000https://yanirseroussi.com/til/2023/12/18/positioning-is-a-common-problem-for-data-scientists/With the commodification of data scientists, the problem of positioning has become more common: My takeaways from Genevieve Hayes interviewing Jonathan Stark.Transfer learning applies to energy market biddinghttps://yanirseroussi.com/til/2023/12/14/transfer-learning-applies-to-energy-market-bidding/Thu, 14 Dec 2023 00:15:00 +0000https://yanirseroussi.com/til/2023/12/14/transfer-learning-applies-to-energy-market-bidding/An interesting approach to bidding of energy storage assets, showing that training on New York data is transferable to Queensland.Supporting volunteer monitoring of marine biodiversity with modern web and data toolshttps://yanirseroussi.com/2023/11/29/supporting-volunteer-monitoring-of-marine-biodiversity-with-modern-web-and-data-tools/Wed, 29 Nov 2023 02:00:00 +0000https://yanirseroussi.com/2023/11/29/supporting-volunteer-monitoring-of-marine-biodiversity-with-modern-web-and-data-tools/Summarising the work Uri Seroussi and I did to improve Reef Life Survey&rsquo;s Reef Species of the World app.Our Blue Machine is changing, but we are not helplesshttps://yanirseroussi.com/til/2023/11/28/our-blue-machine-is-changing-but-we-are-not-helpless/Tue, 28 Nov 2023 06:40:00 +0000https://yanirseroussi.com/til/2023/11/28/our-blue-machine-is-changing-but-we-are-not-helpless/One of my many highlights from Helen Czerski&rsquo;s Blue Machine.You don't need a proprietary API for static mapshttps://yanirseroussi.com/til/2023/11/21/you-dont-need-a-proprietary-api-for-static-maps/Tue, 21 Nov 2023 06:00:00 +0000https://yanirseroussi.com/til/2023/11/21/you-dont-need-a-proprietary-api-for-static-maps/For many use cases, libraries like cartopy are better than the likes of Mapbox and Google Maps.Lessons from reluctant data engineeringhttps://yanirseroussi.com/2023/10/25/lessons-from-reluctant-data-engineering/Wed, 25 Oct 2023 04:45:00 +0000https://yanirseroussi.com/2023/10/25/lessons-from-reluctant-data-engineering/Video and summary of a talk I gave at DataEngBytes Brisbane on what I learned from doing data engineering as part of every data science role I had.Artificial intelligence was a marketing term all along – just call it automationhttps://yanirseroussi.com/til/2023/10/06/artificial-intelligence-was-a-marketing-term-all-along-just-call-it-automation/Fri, 06 Oct 2023 05:00:00 +0000https://yanirseroussi.com/til/2023/10/06/artificial-intelligence-was-a-marketing-term-all-along-just-call-it-automation/Replacing &lsquo;artificial intelligence&rsquo; with &lsquo;automation&rsquo; is a useful trick for cutting through the hype.The lines between solo consulting and product building are blurryhttps://yanirseroussi.com/til/2023/09/25/the-lines-between-solo-consulting-and-product-building-are-blurry/Mon, 25 Sep 2023 00:00:00 +0000https://yanirseroussi.com/til/2023/09/25/the-lines-between-solo-consulting-and-product-building-are-blurry/It turns out that problems like finding a niche and defining the ideal clients are key to any solo business.Google's Rules of Machine Learning still apply in the age of large language modelshttps://yanirseroussi.com/til/2023/09/21/googles-rules-of-machine-learning-still-apply-in-the-age-of-large-language-models/Thu, 21 Sep 2023 21:30:00 +0000https://yanirseroussi.com/til/2023/09/21/googles-rules-of-machine-learning-still-apply-in-the-age-of-large-language-models/Despite the excitement around large language models, building with machine learning remains an engineering problem with established best practices.My rediscovery of quiet writing on the open webhttps://yanirseroussi.com/2023/08/28/my-rediscovery-of-quiet-writing-on-the-open-web/Mon, 28 Aug 2023 05:30:00 +0000https://yanirseroussi.com/2023/08/28/my-rediscovery-of-quiet-writing-on-the-open-web/Reflections on publishing on this website: Writing publicly to share thoughts and documentation beats chasing views and likes.The Minimalist Entrepreneur is too prescriptive for mehttps://yanirseroussi.com/til/2023/08/21/the-minimalist-entrepreneur-is-too-prescriptive-for-me/Mon, 21 Aug 2023 03:15:00 +0000https://yanirseroussi.com/til/2023/08/21/the-minimalist-entrepreneur-is-too-prescriptive-for-me/While I found the story of Gumroad interesting, The Minimalist Entrepreneur seems to over-generalise from the founder&rsquo;s experience.Revisiting Start Small, Stay Small in 2023 (Chapter 2)https://yanirseroussi.com/til/2023/08/17/revisiting-start-small-stay-small-in-2023-chapter-2/Thu, 17 Aug 2023 07:45:00 +0000https://yanirseroussi.com/til/2023/08/17/revisiting-start-small-stay-small-in-2023-chapter-2/A summary of the second chapter of Rob Walling&rsquo;s Start Small, Stay Small, along with my thoughts &amp; reflections.Revisiting Start Small, Stay Small in 2023 (Chapter 1)https://yanirseroussi.com/til/2023/08/16/revisiting-start-small-stay-small-in-2023-chapter-1/Wed, 16 Aug 2023 05:45:00 +0000https://yanirseroussi.com/til/2023/08/16/revisiting-start-small-stay-small-in-2023-chapter-1/A summary of the first chapter of Rob Walling&rsquo;s Start Small, Stay Small, along with my thoughts &amp; reflections.Email notifications on public GitHub commitshttps://yanirseroussi.com/til/2023/08/14/email-notifications-on-public-github-commits/Mon, 14 Aug 2023 05:15:00 +0000https://yanirseroussi.com/til/2023/08/14/email-notifications-on-public-github-commits/GitHub publishes an Atom feed, which means you can use any RSS reader to follow commits.The rule of thirds can probably be ignoredhttps://yanirseroussi.com/til/2023/08/11/the-rule-of-thirds-can-probably-be-ignored/Fri, 11 Aug 2023 03:15:00 +0000https://yanirseroussi.com/til/2023/08/11/the-rule-of-thirds-can-probably-be-ignored/Turns out that the rule of thirds for composing visuals may not be that important.Using YubiKey for SSH accesshttps://yanirseroussi.com/til/2023/07/23/using-yubikey-for-ssh-access/Sun, 23 Jul 2023 00:07:15 +0000https://yanirseroussi.com/til/2023/07/23/using-yubikey-for-ssh-access/Some pointers for setting up SSH access with YubiKey on Ubuntu 22.04.Making a TIL section with Hugo and PaperModhttps://yanirseroussi.com/til/2023/07/17/making-a-til-section-with-hugo-and-papermod/Mon, 17 Jul 2023 00:06:15 +0000https://yanirseroussi.com/til/2023/07/17/making-a-til-section-with-hugo-and-papermod/How I added a Today I Learned section to my Hugo site with the PaperMod theme.You can't save timehttps://yanirseroussi.com/til/2023/07/11/you-cant-save-time/Tue, 11 Jul 2023 00:00:00 +0000https://yanirseroussi.com/til/2023/07/11/you-cant-save-time/Time can be spent doing different activities, but it can&rsquo;t be stored and saved for later.Was data science a failure mode of software engineering?https://yanirseroussi.com/2023/06/30/was-data-science-a-failure-mode-of-software-engineering/Fri, 30 Jun 2023 00:06:30 +0000https://yanirseroussi.com/2023/06/30/was-data-science-a-failure-mode-of-software-engineering/Yes, data science projects have suffered from classic software engineering mistakes, but the field is maturing with the rise of new engineering roles.How hackable are automated coding assessments?https://yanirseroussi.com/2023/05/26/how-hackable-are-automated-coding-assessments/Fri, 26 May 2023 00:03:00 +0000https://yanirseroussi.com/2023/05/26/how-hackable-are-automated-coding-assessments/Exploring the hackability of speed-based coding tests, using CodeSignal&rsquo;s Industry Coding Framework as a case study.Remaining relevant as a small language modelhttps://yanirseroussi.com/2023/04/21/remaining-relevant-as-a-small-language-model/Fri, 21 Apr 2023 00:06:30 +0000https://yanirseroussi.com/2023/04/21/remaining-relevant-as-a-small-language-model/Bing Chat recently quipped that humans are small language models. Here are some of my thoughts on how we small language models can remain relevant (for now).ChatGPT is transformative AIhttps://yanirseroussi.com/2022/12/11/chatgpt-is-transformative-ai/Sun, 11 Dec 2022 00:00:00 +0000https://yanirseroussi.com/2022/12/11/chatgpt-is-transformative-ai/My perspective after a week of using ChatGPT: This is a step change in finding distilled information, and it&rsquo;s only the beginning.Causal Machine Learning is off to a good start, despite some issueshttps://yanirseroussi.com/2022/09/12/causal-machine-learning-book-draft-review/Mon, 12 Sep 2022 02:45:00 +0000https://yanirseroussi.com/2022/09/12/causal-machine-learning-book-draft-review/Reviewing the first three chapters of the book Causal Machine Learning by Robert Osazuwa Ness.The mission matters: Moving to climate tech as a data scientisthttps://yanirseroussi.com/2022/06/06/the-mission-matters-moving-to-climate-tech-as-a-data-scientist/Mon, 06 Jun 2022 00:00:00 +0000https://yanirseroussi.com/2022/06/06/the-mission-matters-moving-to-climate-tech-as-a-data-scientist/Discussing my recent career move into climate tech as a way of doing more to help mitigate dangerous climate change.Building useful machine learning tools keeps getting easier: A fish ID case studyhttps://yanirseroussi.com/2022/03/20/building-useful-machine-learning-tools-keeps-getting-easier-a-fish-id-case-study/Sun, 20 Mar 2022 04:30:00 +0000https://yanirseroussi.com/2022/03/20/building-useful-machine-learning-tools-keeps-getting-easier-a-fish-id-case-study/Lessons learned building a fish ID web app with fast.ai and Streamlit, in an attempt to reduce my fear of missing out on the latest deep learning developments.Analysis strategies in online A/B experiments: Intention-to-treat, per-protocol, and other lessons from clinical trialshttps://yanirseroussi.com/2022/01/14/analysis-strategies-in-online-a-b-experiments/Fri, 14 Jan 2022 00:05:40 +0000https://yanirseroussi.com/2022/01/14/analysis-strategies-in-online-a-b-experiments/Epidemiologists analyse clinical trials to estimate the intention-to-treat and per-protocol effects. This post applies their strategies to online experiments.Use your human brain to avoid artificial intelligence disastershttps://yanirseroussi.com/2021/11/22/use-your-human-brain-to-avoid-artificial-intelligence-disasters/Mon, 22 Nov 2021 03:45:00 +0000https://yanirseroussi.com/2021/11/22/use-your-human-brain-to-avoid-artificial-intelligence-disasters/Overview of a talk I gave at a deep learning course, focusing on AI ethics as the need for humans to think on the context and consequences of applying AI.Migrating from WordPress.com to Hugo on GitHub + Cloudflarehttps://yanirseroussi.com/2021/11/10/migrating-from-wordpress-com-to-hugo-on-github-cloudflare/Wed, 10 Nov 2021 06:30:00 +0000https://yanirseroussi.com/2021/11/10/migrating-from-wordpress-com-to-hugo-on-github-cloudflare/My reasons for switching from WordPress.com to Hugo on GitHub + Cloudflare, along with a summary of the solution components and migration process.My work with Automattichttps://yanirseroussi.com/2021/10/07/my-work-with-automattic/Thu, 07 Oct 2021 00:00:00 +0000https://yanirseroussi.com/2021/10/07/my-work-with-automattic/Back-dated meta-post that gathers my posts on Automattic blogs into a summary of the work I&rsquo;ve done with the company.Some highlights from 2020https://yanirseroussi.com/2021/04/05/some-highlights-from-2020/Mon, 05 Apr 2021 06:41:48 +0000https://yanirseroussi.com/2021/04/05/some-highlights-from-2020/Sharing remote teamwork insights, my climate &amp; sustainability activism, Reef Life Survey publications, and progress on Automattic&rsquo;s Experimentation Platform.Many is not enough: Counting simulations to bootstrap the right wayhttps://yanirseroussi.com/2020/08/24/many-is-not-enough-counting-simulations-to-bootstrap-the-right-way/Mon, 24 Aug 2020 01:35:17 +0000https://yanirseroussi.com/2020/08/24/many-is-not-enough-counting-simulations-to-bootstrap-the-right-way/Going deeper into correct testing of different methods for bootstrap estimation of confidence intervals.Software commodities are eating interesting data science workhttps://yanirseroussi.com/2020/01/11/software-commodities-are-eating-interesting-data-science-work/Sat, 11 Jan 2020 09:22:35 +0000https://yanirseroussi.com/2020/01/11/software-commodities-are-eating-interesting-data-science-work/Being a data scientist can sometimes feel like a race against software commodities that replace interesting work. What can one do to remain relevant?A day in the life of a remote data scientisthttps://yanirseroussi.com/2019/12/12/a-day-in-the-life-of-a-remote-data-scientist/Wed, 11 Dec 2019 22:06:19 +0000https://yanirseroussi.com/2019/12/12/a-day-in-the-life-of-a-remote-data-scientist/Video of a talk I gave on remote data science work at the Data Science Sydney meetup.Bootstrapping the right way?https://yanirseroussi.com/2019/10/06/bootstrapping-the-right-way/Sun, 06 Oct 2019 06:48:07 +0000https://yanirseroussi.com/2019/10/06/bootstrapping-the-right-way/Video and summary of a talk I gave at YOW! Data on bootstrap estimation of confidence intervals.Hackers beware: Bootstrap sampling may be harmfulhttps://yanirseroussi.com/2019/01/08/hackers-beware-bootstrap-sampling-may-be-harmful/Mon, 07 Jan 2019 21:07:56 +0000https://yanirseroussi.com/2019/01/08/hackers-beware-bootstrap-sampling-may-be-harmful/Bootstrap sampling has been promoted as an easy way of modelling uncertainty to hackers without much statistical knowledge. But things aren&rsquo;t that simple.The most practical causal inference book I’ve read (is still a draft)https://yanirseroussi.com/2018/12/24/the-most-practical-causal-inference-book-ive-read-is-still-a-draft/Mon, 24 Dec 2018 02:37:50 +0000https://yanirseroussi.com/2018/12/24/the-most-practical-causal-inference-book-ive-read-is-still-a-draft/Causal Inference by Miguel Hernán and Jamie Robins is a must-read for anyone interested in the area.Reflections on remote data science workhttps://yanirseroussi.com/2018/11/03/reflections-on-remote-data-science-work/Sat, 03 Nov 2018 06:33:13 +0000https://yanirseroussi.com/2018/11/03/reflections-on-remote-data-science-work/Discussing the pluses and minuses of remote work eighteen months after joining Automattic as a data scientist.Defining data science in 2018https://yanirseroussi.com/2018/07/22/defining-data-science-in-2018/Sun, 22 Jul 2018 08:27:43 +0000https://yanirseroussi.com/2018/07/22/defining-data-science-in-2018/Updating my definition of data science to match changes in the field. It is now broader than before, but its ultimate goal is still to support decisions.Advice for aspiring data scientists and other FAQshttps://yanirseroussi.com/2017/10/15/advice-for-aspiring-data-scientists-and-other-faqs/Sun, 15 Oct 2017 09:15:25 +0000https://yanirseroussi.com/2017/10/15/advice-for-aspiring-data-scientists-and-other-faqs/Frequently asked questions by visitors to this site, especially around entering the data science field.State of Bandcamp Recommender, Late 2017https://yanirseroussi.com/2017/09/02/state-of-bandcamp-recommender/Sat, 02 Sep 2017 10:19:02 +0000https://yanirseroussi.com/2017/09/02/state-of-bandcamp-recommender/Call for BCRecommender maintainers followed by a decision to shut it down, as I don&rsquo;t have enough time and Bandcamp now offers recommendations.My 10-step path to becoming a remote data scientist with Automattichttps://yanirseroussi.com/2017/07/29/my-10-step-path-to-becoming-a-remote-data-scientist-with-automattic/Sat, 29 Jul 2017 05:39:26 +0000https://yanirseroussi.com/2017/07/29/my-10-step-path-to-becoming-a-remote-data-scientist-with-automattic/I wanted a well-paid data science-y remote job with an established company that offers a good life balance and makes products I care about. I got it eventually.Exploring and visualising Reef Life Survey datahttps://yanirseroussi.com/2017/06/03/exploring-and-visualising-reef-life-survey-data/Sat, 03 Jun 2017 00:49:05 +0000https://yanirseroussi.com/2017/06/03/exploring-and-visualising-reef-life-survey-data/Web tools I built to visualise Reef Life Survey data and assist citizen scientists in underwater visual census work.Customer lifetime value and the proliferation of misinformation on the internethttps://yanirseroussi.com/2017/01/08/customer-lifetime-value-and-the-proliferation-of-misinformation-on-the-internet/Sun, 08 Jan 2017 20:02:30 +0000https://yanirseroussi.com/2017/01/08/customer-lifetime-value-and-the-proliferation-of-misinformation-on-the-internet/There&rsquo;s a lot of misleading content on the estimation of customer lifetime value. Here&rsquo;s what I learned about doing it well.Ask Why! Finding motives, causes, and purpose in data sciencehttps://yanirseroussi.com/2016/09/19/ask-why-finding-motives-causes-and-purpose-in-data-science/Mon, 19 Sep 2016 21:28:44 +0000https://yanirseroussi.com/2016/09/19/ask-why-finding-motives-causes-and-purpose-in-data-science/Video and summary of a talk I gave at the Data Science Sydney meetup, about going beyond the what &amp; how of predictive modelling.If you don’t pay attention, data can drive you off a cliffhttps://yanirseroussi.com/2016/08/21/seven-ways-to-be-data-driven-off-a-cliff/Sun, 21 Aug 2016 21:34:17 +0000https://yanirseroussi.com/2016/08/21/seven-ways-to-be-data-driven-off-a-cliff/Seven common mistakes to avoid when working with data, such as ignoring uncertainty and confusing observed and unobserved quantities.Is Data Scientist a useless job title?https://yanirseroussi.com/2016/08/04/is-data-scientist-a-useless-job-title/Thu, 04 Aug 2016 22:26:03 +0000https://yanirseroussi.com/2016/08/04/is-data-scientist-a-useless-job-title/It seems like anyone who touches data can call themselves a data scientist, which makes the title useless. The work they do can still be useful, though.Making Bayesian A/B testing more accessiblehttps://yanirseroussi.com/2016/06/19/making-bayesian-ab-testing-more-accessible/Sun, 19 Jun 2016 10:32:15 +0000https://yanirseroussi.com/2016/06/19/making-bayesian-ab-testing-more-accessible/A web tool I built to interpret A/B test results in a Bayesian way, including prior specification, visualisations, and decision rules.Diving deeper into causality: Pearl, Kleinberg, Hill, and untested assumptionshttps://yanirseroussi.com/2016/05/15/diving-deeper-into-causality-pearl-kleinberg-hill-and-untested-assumptions/Sat, 14 May 2016 19:57:03 +0000https://yanirseroussi.com/2016/05/15/diving-deeper-into-causality-pearl-kleinberg-hill-and-untested-assumptions/Discussing the need for untested assumptions and temporality in causal inference. Mostly based on Samantha Kleinberg&rsquo;s Causality, Probability, and Time.The rise of greedy robotshttps://yanirseroussi.com/2016/03/20/the-rise-of-greedy-robots/Sun, 20 Mar 2016 20:33:43 +0000https://yanirseroussi.com/2016/03/20/the-rise-of-greedy-robots/Is artificial/machine intelligence a future threat? I argue that it&rsquo;s already here, with greedy robots already dominating our lives.Why you should stop worrying about deep learning and deepen your understanding of causality insteadhttps://yanirseroussi.com/2016/02/14/why-you-should-stop-worrying-about-deep-learning-and-deepen-your-understanding-of-causality-instead/Sun, 14 Feb 2016 11:04:11 +0000https://yanirseroussi.com/2016/02/14/why-you-should-stop-worrying-about-deep-learning-and-deepen-your-understanding-of-causality-instead/Causality is often overlooked but is of much higher relevance to most data scientists than deep learning.The joys of offline data collectionhttps://yanirseroussi.com/2016/01/24/the-joys-of-offline-data-collection/Sun, 24 Jan 2016 00:32:25 +0000https://yanirseroussi.com/2016/01/24/the-joys-of-offline-data-collection/Insights on data collection and machine learning from spending a month sailing, diving, and counting fish with Reef Life Survey.This holiday season, give me real insightshttps://yanirseroussi.com/2015/12/08/this-holiday-season-give-me-real-insights/Tue, 08 Dec 2015 06:57:25 +0000https://yanirseroussi.com/2015/12/08/this-holiday-season-give-me-real-insights/Some companies present raw data or information as &ldquo;insights&rdquo;. This post surveys some examples, and discusses how they can be turned into real insights.The hardest parts of data sciencehttps://yanirseroussi.com/2015/11/23/the-hardest-parts-of-data-science/Mon, 23 Nov 2015 04:14:21 +0000https://yanirseroussi.com/2015/11/23/the-hardest-parts-of-data-science/Defining feasible problems and coming up with reasonable ways of measuring solutions is harder than building accurate models or obtaining clean data.Migrating a simple web application from MongoDB to Elasticsearchhttps://yanirseroussi.com/2015/11/04/migrating-a-simple-web-application-from-mongodb-to-elasticsearch/Wed, 04 Nov 2015 03:53:18 +0000https://yanirseroussi.com/2015/11/04/migrating-a-simple-web-application-from-mongodb-to-elasticsearch/Migrating BCRecommender from MongoDB to Elasticsearch made it possible to offer a richer search experience to users at a similar cost, among other benefits.Miscommunicating science: Simplistic models, nutritionism, and the art of storytellinghttps://yanirseroussi.com/2015/10/19/nutritionism-and-the-need-for-complex-models-to-explain-complex-phenomena/Mon, 19 Oct 2015 00:02:32 +0000https://yanirseroussi.com/2015/10/19/nutritionism-and-the-need-for-complex-models-to-explain-complex-phenomena/Nutritionism is a special case of misinterpretation and miscommunication of scientific results – something many data scientists encounter in their work.The wonderful world of recommender systemshttps://yanirseroussi.com/2015/10/02/the-wonderful-world-of-recommender-systems/Fri, 02 Oct 2015 05:25:57 +0000https://yanirseroussi.com/2015/10/02/the-wonderful-world-of-recommender-systems/Giving an overview of the field and common paradigms, and debunking five common myths about recommender systems.You don’t need a data scientist (yet)https://yanirseroussi.com/2015/08/24/you-dont-need-a-data-scientist-yet/Mon, 24 Aug 2015 08:25:30 +0000https://yanirseroussi.com/2015/08/24/you-dont-need-a-data-scientist-yet/Hiring data scientists prematurely is wasteful and frustrating. Here are some questions to ask before you hire your first data scientist.Goodbye, Parse.comhttps://yanirseroussi.com/2015/07/31/goodbye-parse-com/Fri, 31 Jul 2015 03:29:50 +0000https://yanirseroussi.com/2015/07/31/goodbye-parse-com/Migrating my web apps away from Parse.com due to reliability issues. Self-hosting is a better solution.Learning about deep learning through album cover classificationhttps://yanirseroussi.com/2015/07/06/learning-about-deep-learning-through-album-cover-classification/Mon, 06 Jul 2015 22:21:42 +0000https://yanirseroussi.com/2015/07/06/learning-about-deep-learning-through-album-cover-classification/Progress on my album cover classification project, highlighting lessons that would be useful to others who are getting started with deep learning.Deep learning resourceshttps://yanirseroussi.com/deep-learning-resources/Mon, 06 Jul 2015 00:38:44 +0000https://yanirseroussi.com/deep-learning-resources/This page summarises the deep learning resources I&rsquo;ve consulted in my album cover classification project. Tutorials and blog posts Convolutional Neural Networks for Visual Recognition Stanford course notes: an excellent resource, very up-to-date and useful, despite still being a work in progress DeepLearning.net&rsquo;s Theano-based tutorials: not as up-to-date as the Stanford course notes, but still a good introduction to some of the theory and general Theano usage Lasagne&rsquo;s documentation and tutorials: still a bit lacking, but good when you know what you&rsquo;re looking for lasagne4newbs: Lasagne&rsquo;s convnet example with richer comments Using convolutional neural nets to detect facial keypoints tutorial: the resource that made me want to use Lasagne Classifying plankton with deep neural networks: an epic post, which I found while looking for Lasagne examples Various Wikipedia pages: a bit disappointing – the above resources are much better Papers Adam: a method for stochastic optimization (Kingma and Ba, 2015): an improvement over SGD with Nesterov momentum, AdaGrad and RMSProp, which I found to be useful in practice Algorithms for Hyper-Parameter Optimization (Bergstra et al.Hopping on the deep learning bandwagonhttps://yanirseroussi.com/2015/06/06/hopping-on-the-deep-learning-bandwagon/Sat, 06 Jun 2015 05:00:22 +0000https://yanirseroussi.com/2015/06/06/hopping-on-the-deep-learning-bandwagon/To become proficient at solving data science problems, you need to get your hands dirty. Here, I used album cover classification to learn about deep learning.First steps in data science: author-aware sentiment analysishttps://yanirseroussi.com/2015/05/02/first-steps-in-data-science-author-aware-sentiment-analysis/Sat, 02 May 2015 08:31:10 +0000https://yanirseroussi.com/2015/05/02/first-steps-in-data-science-author-aware-sentiment-analysis/I became a data scientist by doing a PhD, but the same steps can be followed without a formal education program.My divestment from fossil fuelshttps://yanirseroussi.com/2015/04/24/my-divestment-from-fossil-fuels/Fri, 24 Apr 2015 00:19:36 +0000https://yanirseroussi.com/2015/04/24/my-divestment-from-fossil-fuels/Recent choices I&rsquo;ve made to reduce my exposure to fossil fuels, including practical steps that can be taken by Australians and generally applicable lessons.My PhD workhttps://yanirseroussi.com/phd-work/Mon, 30 Mar 2015 03:23:33 +0000https://yanirseroussi.com/phd-work/An overview of my PhD in data science / artificial intelligence. Thesis title: Text Mining and Rating Prediction with Topical User Models.The long road to a lifestyle businesshttps://yanirseroussi.com/2015/03/22/the-long-road-to-a-lifestyle-business/Sun, 22 Mar 2015 09:43:47 +0000https://yanirseroussi.com/2015/03/22/the-long-road-to-a-lifestyle-business/Progress since leaving my last full-time job and setting on an independent path that includes data science consulting and work on my own projects.Learning to rank for personalised search (Yandex Search Personalisation – Kaggle Competition Summary – Part 2)https://yanirseroussi.com/2015/02/11/learning-to-rank-for-personalised-search-yandex-search-personalisation-kaggle-competition-summary-part-2/Wed, 11 Feb 2015 06:34:17 +0000https://yanirseroussi.com/2015/02/11/learning-to-rank-for-personalised-search-yandex-search-personalisation-kaggle-competition-summary-part-2/My team&rsquo;s solution to the Yandex Search Personalisation competition (finished 9th out of 194 teams).Is thinking like a search engine possible? (Yandex search personalisation – Kaggle competition summary – part 1)https://yanirseroussi.com/2015/01/29/is-thinking-like-a-search-engine-possible-yandex-search-personalisation-kaggle-competition-summary-part-1/Thu, 29 Jan 2015 10:37:39 +0000https://yanirseroussi.com/2015/01/29/is-thinking-like-a-search-engine-possible-yandex-search-personalisation-kaggle-competition-summary-part-1/Insights on search personalisation and SEO from participating in a Kaggle competition (finished 9th out of 194 teams).Automating Parse.com bulk data importshttps://yanirseroussi.com/2015/01/15/automating-parse-com-bulk-data-imports/Thu, 15 Jan 2015 04:41:16 +0000https://yanirseroussi.com/2015/01/15/automating-parse-com-bulk-data-imports/A script for importing data into the Parse backend-as-a-service.Stochastic Gradient Boosting: Choosing the Best Number of Iterationshttps://yanirseroussi.com/2014/12/29/stochastic-gradient-boosting-choosing-the-best-number-of-iterations/Mon, 29 Dec 2014 02:30:06 +0000https://yanirseroussi.com/2014/12/29/stochastic-gradient-boosting-choosing-the-best-number-of-iterations/Exploring an approach to choosing the optimal number of iterations in stochastic gradient boosting, following a bug I found in scikit-learn.SEO: Mostly about showing up?https://yanirseroussi.com/2014/12/15/seo-mostly-about-showing-up/Mon, 15 Dec 2014 04:25:25 +0000https://yanirseroussi.com/2014/12/15/seo-mostly-about-showing-up/Increasing SEO traffic to BCRecommender by adding content and opening up more pages for crawling. It turns out that thin content is better than no content.Fitting noise: Forecasting the sale price of bulldozers (Kaggle competition summary)https://yanirseroussi.com/2014/11/19/fitting-noise-forecasting-the-sale-price-of-bulldozers-kaggle-competition-summary/Wed, 19 Nov 2014 09:17:34 +0000https://yanirseroussi.com/2014/11/19/fitting-noise-forecasting-the-sale-price-of-bulldozers-kaggle-competition-summary/Summary of a Kaggle competition to forecast bulldozer sale price, where I finished 9th out of 476 teams.BCRecommender Traction Updatehttps://yanirseroussi.com/2014/11/05/bcrecommender-traction-update/Wed, 05 Nov 2014 02:29:35 +0000https://yanirseroussi.com/2014/11/05/bcrecommender-traction-update/Update on BCRecommender traction using three channels: blogger outreach, search engine optimisation, and content marketing.What is data science?https://yanirseroussi.com/2014/10/23/what-is-data-science/Thu, 23 Oct 2014 03:22:08 +0000https://yanirseroussi.com/2014/10/23/what-is-data-science/Data science has been a hot term in the past few years. Still, there isn&rsquo;t a single definition of the field. This post discusses my favourite definition.Greek Media Monitoring Kaggle competition: My approachhttps://yanirseroussi.com/2014/10/07/greek-media-monitoring-kaggle-competition-my-approach/Tue, 07 Oct 2014 03:21:35 +0000https://yanirseroussi.com/2014/10/07/greek-media-monitoring-kaggle-competition-my-approach/Summary of my approach to the Greek Media Monitoring Kaggle competition, where I finished 6th out of 120 teams.Applying the Traction Book’s Bullseye framework to BCRecommenderhttps://yanirseroussi.com/2014/09/24/applying-the-traction-books-bullseye-framework-to-bcrecommender/Wed, 24 Sep 2014 04:57:39 +0000https://yanirseroussi.com/2014/09/24/applying-the-traction-books-bullseye-framework-to-bcrecommender/Ranking 19 channels with the goal of getting traction for BCRecommender.Bandcamp recommendation and discovery algorithmshttps://yanirseroussi.com/2014/09/19/bandcamp-recommendation-and-discovery-algorithms/Fri, 19 Sep 2014 14:26:55 +0000https://yanirseroussi.com/2014/09/19/bandcamp-recommendation-and-discovery-algorithms/The recommendation backend for my BCRecommender service for personalised Bandcamp music discovery.Building a recommender system on a shoestring budget (or: BCRecommender part 2 – general system layout)https://yanirseroussi.com/2014/09/07/building-a-recommender-system-on-a-shoestring-budget/Sun, 07 Sep 2014 10:48:44 +0000https://yanirseroussi.com/2014/09/07/building-a-recommender-system-on-a-shoestring-budget/Iterating on my BCRecommender service with the goal of keeping costs low while providing a valuable music recommendation service.Building a Bandcamp recommender system (part 1 – motivation)https://yanirseroussi.com/2014/08/30/building-a-bandcamp-recommender-system-part-1-motivation/Sat, 30 Aug 2014 08:11:38 +0000https://yanirseroussi.com/2014/08/30/building-a-bandcamp-recommender-system-part-1-motivation/My motivation behind building BCRecommender, a free recommendation &amp; discovery service for Bandcamp music.How to (almost) win Kaggle competitionshttps://yanirseroussi.com/2014/08/24/how-to-almost-win-kaggle-competitions/Sun, 24 Aug 2014 12:40:53 +0000https://yanirseroussi.com/2014/08/24/how-to-almost-win-kaggle-competitions/Summary of a talk I gave at the Data Science Sydney meetup with ten tips on almost-winning Kaggle competitions.Data’s hierarchy of needshttps://yanirseroussi.com/2014/08/17/datas-hierarchy-of-needs/Sun, 17 Aug 2014 13:09:30 +0000https://yanirseroussi.com/2014/08/17/datas-hierarchy-of-needs/Discussing the hierarchy of needs proposed by Jay Kreps. Key takeaway: Data-driven algorithms &amp; insights can only be as good as the underlying data.Kaggle competition tips and summarieshttps://yanirseroussi.com/kaggle/Sat, 05 Apr 2014 23:46:10 +0000https://yanirseroussi.com/kaggle/Pointers to all my Kaggle advice posts and competition summaries.Kaggle beginner tipshttps://yanirseroussi.com/2014/01/19/kaggle-beginner-tips/Sun, 19 Jan 2014 10:34:28 +0000https://yanirseroussi.com/2014/01/19/kaggle-beginner-tips/First post! An email I sent to members of the Data Science Sydney Meetup with tips on how to get started with Kaggle competitions.About Mehttps://yanirseroussi.com/about/Mon, 01 Jan 0001 00:00:00 +0000https://yanirseroussi.com/about/About Yanir Seroussi, a hands-on data tech lead with over a decade of experience.Book a free fifteen-minute callhttps://yanirseroussi.com/free-intro-call/Mon, 01 Jan 0001 00:00:00 +0000https://yanirseroussi.com/free-intro-call/Booking form for a quick intro call with Yanir Seroussi.Causal inference resourceshttps://yanirseroussi.com/causal-inference-resources/Mon, 01 Jan 0001 00:00:00 +0000https://yanirseroussi.com/causal-inference-resources/This is a list of some causal inference resources, which I update from time to time. You can also check out my posts on causal inference and A/B testing. Books: Causal Inference: What if by Miguel Hernán and Jamie Robins: The most practical book I&rsquo;ve read. Highly recommended. Trustworthy Online Controlled Experiments : A Practical Guide to A/B Testing by Ron Kohavi, Diane Tang, and Ya Xu: Building on the authors&rsquo; decades of industry experience, this is pretty much the bible of online experiments, which is how causal inference is often done in practice.Data & AI Consulting for Startupshttps://yanirseroussi.com/consult/Mon, 01 Jan 0001 00:00:00 +0000https://yanirseroussi.com/consult/Yanir Seroussi&rsquo;s Data &amp; AI consulting services, mostly targeting startups and scaleups focused on positive-impact outcomes.Free Guide: Data-to-AI Health Check for Startupshttps://yanirseroussi.com/data-to-ai-health-check/Mon, 01 Jan 0001 00:00:00 +0000https://yanirseroussi.com/data-to-ai-health-check/Download a free PDF guide that helps you assess a startup&rsquo;s Data-to-AI health by probing eight key areas.Stay in touchhttps://yanirseroussi.com/contact/Mon, 01 Jan 0001 00:00:00 +0000https://yanirseroussi.com/contact/Contact me or subscribe to the mailing list.Talkshttps://yanirseroussi.com/talks/Mon, 01 Jan 0001 00:00:00 +0000https://yanirseroussi.com/talks/Just a list of some talks I&rsquo;ve given, saved here for future reference and for general public benefit. diff --git a/posts/index.html b/posts/index.html index 41704bf04..07c4ce3af 100644 --- a/posts/index.html +++ b/posts/index.html @@ -11,7 +11,7 @@ ">

    Browse Posts

    Browse my main posts in reverse chronological order below, or by tag. You may also want to check out my shorter-form TIL (today I learned) posts.
      -
      Alternatively, subscribe to RSS feed.

      How to avoid startups with poor development processes

      Questions that prospective data specialists and engineers should ask about development processes before accepting a startup role.

      June 3, 2024

      Plumbing, Decisions, and Automation: De-hyping Data & AI

      Three essential questions to understand where an organisation stands when it comes to Data & AI (with zero hype).

      May 27, 2024

      Question startup culture before accepting a data-to-AI role

      Eight questions that prospective data-to-AI employees should ask about a startup’s work and data culture.

      May 20, 2024

      Probing the People aspects of an early-stage startup

      Ten questions that prospective employees should ask about a startup’s team, especially for data-centric roles.

      May 13, 2024

      Business questions to ask before taking a startup data role

      Fourteen questions that prospective employees should ask about a startup’s business model and product, especially for data-focused roles.

      May 6, 2024

      Mentorship and the art of actionable advice

      Reflections on what it takes to package expertise and deliver timely, actionable advice outside the context of employee relationships.

      April 29, 2024

      Assessing a startup's data-to-AI health

      Reviewing the areas that should be assessed to determine a startup’s opportunities and challenges on the data/AI/ML front.

      April 22, 2024

      AI does not obviate the need for testing and observability

      It’s easy to prototype with AI, but production-grade AI apps require even more thorough testing and observability than traditional software.

      April 15, 2024

      My experience as a Data Tech Lead with Work on Climate

      The story of how I joined Work on Climate as a volunteer and became its data tech lead, with lessons applied to consulting & fractional work.

      April 8, 2024

      Artificial intelligence, automation, and the art of counting fish

      Discussing the use of AI to automate underwater marine surveys as an example of the uneven distribution of technological advancement.

      April 1, 2024

      Questions to consider when using AI for PDF data extraction

      Discussing considerations that arise when attempting to automate the extraction of structured data from PDFs and similar documents.

      March 11, 2024

      Two types of startup data problems

      Classifying startups as ML-centric or non-ML is a helpful exercise to uncover the data challenges they’re likely to face.

      March 4, 2024

      Avoiding AI complexity: First, write no code

      Two stories of getting AI functionality to production, which demonstrate the risks inherent in custom development versus starting with a no-code approach.

      February 26, 2024

      Building your startup's minimum viable data stack

      First post in a series on building a minimum viable data stack for startups, introducing key definitions, components, and considerations.

      February 19, 2024

      Nudging ChatGPT to invent books you have no time to read

      Getting ChatGPT Plus to elaborate on possible book content and produce a PDF cheatsheet, with the goal of learning about its capabilities.

      February 12, 2024

      Substance over titles: Your first data hire may be a data scientist

      Advice for hiring a startup’s first data person: match skills to business needs, consider contractors, and get help from data people.

      February 5, 2024

      New decade, new tagline: Data & AI for Impact

      Shifting focus to ‘Data & AI for Impact’, with more startup-related content, increased posting frequency, and deeper audience engagement.

      January 19, 2024

      Supporting volunteer monitoring of marine biodiversity with modern web and data tools

      Summarising the work Uri Seroussi and I did to improve Reef Life Survey’s Reef Species of the World app.

      November 29, 2023

      Lessons from reluctant data engineering

      Video and summary of a talk I gave at DataEngBytes Brisbane on what I learned from doing data engineering as part of every data science role I had.

      October 25, 2023

      My rediscovery of quiet writing on the open web

      Reflections on publishing on this website: Writing publicly to share thoughts and documentation beats chasing views and likes.

      August 28, 2023

      Was data science a failure mode of software engineering?

      Yes, data science projects have suffered from classic software engineering mistakes, but the field is maturing with the rise of new engineering roles.

      June 30, 2023

      How hackable are automated coding assessments?

      Exploring the hackability of speed-based coding tests, using CodeSignal’s Industry Coding Framework as a case study.

      May 26, 2023

      Remaining relevant as a small language model

      Bing Chat recently quipped that humans are small language models. Here are some of my thoughts on how we small language models can remain relevant (for now).

      April 21, 2023

      ChatGPT is transformative AI

      My perspective after a week of using ChatGPT: This is a step change in finding distilled information, and it’s only the beginning.

      December 11, 2022

      Causal Machine Learning is off to a good start, despite some issues

      Reviewing the first three chapters of the book Causal Machine Learning by Robert Osazuwa Ness.

      September 12, 2022

      The mission matters: Moving to climate tech as a data scientist

      Discussing my recent career move into climate tech as a way of doing more to help mitigate dangerous climate change.

      June 6, 2022

      Building useful machine learning tools keeps getting easier: A fish ID case study

      Lessons learned building a fish ID web app with fast.ai and Streamlit, in an attempt to reduce my fear of missing out on the latest deep learning developments.

      March 20, 2022

      Analysis strategies in online A/B experiments: Intention-to-treat, per-protocol, and other lessons from clinical trials

      Epidemiologists analyse clinical trials to estimate the intention-to-treat and per-protocol effects. This post applies their strategies to online experiments.

      January 14, 2022

      Use your human brain to avoid artificial intelligence disasters

      Overview of a talk I gave at a deep learning course, focusing on AI ethics as the need for humans to think on the context and consequences of applying AI.

      November 22, 2021

      Migrating from WordPress.com to Hugo on GitHub + Cloudflare

      My reasons for switching from WordPress.com to Hugo on GitHub + Cloudflare, along with a summary of the solution components and migration process.

      November 10, 2021

      My work with Automattic

      Back-dated meta-post that gathers my posts on Automattic blogs into a summary of the work I’ve done with the company.

      October 7, 2021

      Some highlights from 2020

      Sharing remote teamwork insights, my climate & sustainability activism, Reef Life Survey publications, and progress on Automattic’s Experimentation Platform.

      April 5, 2021

      Many is not enough: Counting simulations to bootstrap the right way

      Going deeper into correct testing of different methods for bootstrap estimation of confidence intervals.

      August 24, 2020

      Software commodities are eating interesting data science work

      Being a data scientist can sometimes feel like a race against software commodities that replace interesting work. What can one do to remain relevant?

      January 11, 2020

      A day in the life of a remote data scientist

      Video of a talk I gave on remote data science work at the Data Science Sydney meetup.

      December 11, 2019

      Bootstrapping the right way?

      Video and summary of a talk I gave at YOW! Data on bootstrap estimation of confidence intervals.

      October 6, 2019

      Hackers beware: Bootstrap sampling may be harmful

      Bootstrap sampling has been promoted as an easy way of modelling uncertainty to hackers without much statistical knowledge. But things aren’t that simple.

      January 7, 2019

      The most practical causal inference book I’ve read (is still a draft)

      Causal Inference by Miguel Hernán and Jamie Robins is a must-read for anyone interested in the area.

      December 24, 2018

      Reflections on remote data science work

      Discussing the pluses and minuses of remote work eighteen months after joining Automattic as a data scientist.

      November 3, 2018

      Defining data science in 2018

      Updating my definition of data science to match changes in the field. It is now broader than before, but its ultimate goal is still to support decisions.

      July 22, 2018

      Advice for aspiring data scientists and other FAQs

      Frequently asked questions by visitors to this site, especially around entering the data science field.

      October 15, 2017

      State of Bandcamp Recommender, Late 2017

      Call for BCRecommender maintainers followed by a decision to shut it down, as I don’t have enough time and Bandcamp now offers recommendations.

      September 2, 2017

      My 10-step path to becoming a remote data scientist with Automattic

      I wanted a well-paid data science-y remote job with an established company that offers a good life balance and makes products I care about. I got it eventually.

      July 29, 2017

      Exploring and visualising Reef Life Survey data

      Web tools I built to visualise Reef Life Survey data and assist citizen scientists in underwater visual census work.

      June 3, 2017

      Customer lifetime value and the proliferation of misinformation on the internet

      There’s a lot of misleading content on the estimation of customer lifetime value. Here’s what I learned about doing it well.

      January 8, 2017

      Ask Why! Finding motives, causes, and purpose in data science

      Video and summary of a talk I gave at the Data Science Sydney meetup, about going beyond the what & how of predictive modelling.

      September 19, 2016

      If you don’t pay attention, data can drive you off a cliff

      Seven common mistakes to avoid when working with data, such as ignoring uncertainty and confusing observed and unobserved quantities.

      August 21, 2016

      Is Data Scientist a useless job title?

      It seems like anyone who touches data can call themselves a data scientist, which makes the title useless. The work they do can still be useful, though.

      August 4, 2016

      Making Bayesian A/B testing more accessible

      A web tool I built to interpret A/B test results in a Bayesian way, including prior specification, visualisations, and decision rules.

      June 19, 2016

      Diving deeper into causality: Pearl, Kleinberg, Hill, and untested assumptions

      Discussing the need for untested assumptions and temporality in causal inference. Mostly based on Samantha Kleinberg’s Causality, Probability, and Time.

      May 14, 2016

      The rise of greedy robots

      Is artificial/machine intelligence a future threat? I argue that it’s already here, with greedy robots already dominating our lives.

      March 20, 2016

      Why you should stop worrying about deep learning and deepen your understanding of causality instead

      Causality is often overlooked but is of much higher relevance to most data scientists than deep learning.

      February 14, 2016

      The joys of offline data collection

      Insights on data collection and machine learning from spending a month sailing, diving, and counting fish with Reef Life Survey.

      January 24, 2016

      This holiday season, give me real insights

      Some companies present raw data or information as “insights”. This post surveys some examples, and discusses how they can be turned into real insights.

      December 8, 2015

      The hardest parts of data science

      Defining feasible problems and coming up with reasonable ways of measuring solutions is harder than building accurate models or obtaining clean data.

      November 23, 2015

      Migrating a simple web application from MongoDB to Elasticsearch

      Migrating BCRecommender from MongoDB to Elasticsearch made it possible to offer a richer search experience to users at a similar cost, among other benefits.

      November 4, 2015

      Miscommunicating science: Simplistic models, nutritionism, and the art of storytelling

      Nutritionism is a special case of misinterpretation and miscommunication of scientific results – something many data scientists encounter in their work.

      October 19, 2015

      The wonderful world of recommender systems

      Giving an overview of the field and common paradigms, and debunking five common myths about recommender systems.

      October 2, 2015

      You don’t need a data scientist (yet)

      Hiring data scientists prematurely is wasteful and frustrating. Here are some questions to ask before you hire your first data scientist.

      August 24, 2015

      Goodbye, Parse.com

      Migrating my web apps away from Parse.com due to reliability issues. Self-hosting is a better solution.

      July 31, 2015

      Learning about deep learning through album cover classification

      Progress on my album cover classification project, highlighting lessons that would be useful to others who are getting started with deep learning.

      July 6, 2015

      Hopping on the deep learning bandwagon

      To become proficient at solving data science problems, you need to get your hands dirty. Here, I used album cover classification to learn about deep learning.

      June 6, 2015

      First steps in data science: author-aware sentiment analysis

      I became a data scientist by doing a PhD, but the same steps can be followed without a formal education program.

      May 2, 2015

      My divestment from fossil fuels

      Recent choices I’ve made to reduce my exposure to fossil fuels, including practical steps that can be taken by Australians and generally applicable lessons.

      April 24, 2015

      My PhD work

      An overview of my PhD in data science / artificial intelligence. Thesis title: Text Mining and Rating Prediction with Topical User Models.

      March 30, 2015

      The long road to a lifestyle business

      Progress since leaving my last full-time job and setting on an independent path that includes data science consulting and work on my own projects.

      March 22, 2015

      Learning to rank for personalised search (Yandex Search Personalisation – Kaggle Competition Summary – Part 2)

      My team’s solution to the Yandex Search Personalisation competition (finished 9th out of 194 teams).

      February 11, 2015

      Is thinking like a search engine possible? (Yandex search personalisation – Kaggle competition summary – part 1)

      Insights on search personalisation and SEO from participating in a Kaggle competition (finished 9th out of 194 teams).

      January 29, 2015

      Automating Parse.com bulk data imports

      A script for importing data into the Parse backend-as-a-service.

      January 15, 2015

      Stochastic Gradient Boosting: Choosing the Best Number of Iterations

      Exploring an approach to choosing the optimal number of iterations in stochastic gradient boosting, following a bug I found in scikit-learn.

      December 29, 2014

      SEO: Mostly about showing up?

      Increasing SEO traffic to BCRecommender by adding content and opening up more pages for crawling. It turns out that thin content is better than no content.

      December 15, 2014

      Fitting noise: Forecasting the sale price of bulldozers (Kaggle competition summary)

      Summary of a Kaggle competition to forecast bulldozer sale price, where I finished 9th out of 476 teams.

      November 19, 2014

      BCRecommender Traction Update

      Update on BCRecommender traction using three channels: blogger outreach, search engine optimisation, and content marketing.

      November 5, 2014

      What is data science?

      Data science has been a hot term in the past few years. Still, there isn’t a single definition of the field. This post discusses my favourite definition.

      October 23, 2014

      Greek Media Monitoring Kaggle competition: My approach

      Summary of my approach to the Greek Media Monitoring Kaggle competition, where I finished 6th out of 120 teams.

      October 7, 2014

      Applying the Traction Book’s Bullseye framework to BCRecommender

      Ranking 19 channels with the goal of getting traction for BCRecommender.

      September 24, 2014

      Bandcamp recommendation and discovery algorithms

      The recommendation backend for my BCRecommender service for personalised Bandcamp music discovery.

      September 19, 2014

      Building a recommender system on a shoestring budget (or: BCRecommender part 2 – general system layout)

      Iterating on my BCRecommender service with the goal of keeping costs low while providing a valuable music recommendation service.

      September 7, 2014

      Building a Bandcamp recommender system (part 1 – motivation)

      My motivation behind building BCRecommender, a free recommendation & discovery service for Bandcamp music.

      August 30, 2014

      How to (almost) win Kaggle competitions

      Summary of a talk I gave at the Data Science Sydney meetup with ten tips on almost-winning Kaggle competitions.

      August 24, 2014

      Data’s hierarchy of needs

      Discussing the hierarchy of needs proposed by Jay Kreps. Key takeaway: Data-driven algorithms & insights can only be as good as the underlying data.

      August 17, 2014

      Kaggle competition tips and summaries

      Pointers to all my Kaggle advice posts and competition summaries.

      April 5, 2014

      Kaggle beginner tips

      First post! An email I sent to members of the Data Science Sydney Meetup with tips on how to get started with Kaggle competitions.

      January 19, 2014
      Text and figures licensed under CC BY-NC-ND 4.0 by Yanir Seroussi, except where noted otherwise  | +
      Alternatively, subscribe to RSS feed.

      Startup data health starts with healthy event tracking

      Expanding on the startup health check question of tracking Kukuyeva’s five business aspects as wide events.

      June 10, 2024

      How to avoid startups with poor development processes

      Questions that prospective data specialists and engineers should ask about development processes before accepting a startup role.

      June 3, 2024

      Plumbing, Decisions, and Automation: De-hyping Data & AI

      Three essential questions to understand where an organisation stands when it comes to Data & AI (with zero hype).

      May 27, 2024

      Question startup culture before accepting a data-to-AI role

      Eight questions that prospective data-to-AI employees should ask about a startup’s work and data culture.

      May 20, 2024

      Probing the People aspects of an early-stage startup

      Ten questions that prospective employees should ask about a startup’s team, especially for data-centric roles.

      May 13, 2024

      Business questions to ask before taking a startup data role

      Fourteen questions that prospective employees should ask about a startup’s business model and product, especially for data-focused roles.

      May 6, 2024

      Mentorship and the art of actionable advice

      Reflections on what it takes to package expertise and deliver timely, actionable advice outside the context of employee relationships.

      April 29, 2024

      Assessing a startup's data-to-AI health

      Reviewing the areas that should be assessed to determine a startup’s opportunities and challenges on the data/AI/ML front.

      April 22, 2024

      AI does not obviate the need for testing and observability

      It’s easy to prototype with AI, but production-grade AI apps require even more thorough testing and observability than traditional software.

      April 15, 2024

      My experience as a Data Tech Lead with Work on Climate

      The story of how I joined Work on Climate as a volunteer and became its data tech lead, with lessons applied to consulting & fractional work.

      April 8, 2024

      Artificial intelligence, automation, and the art of counting fish

      Discussing the use of AI to automate underwater marine surveys as an example of the uneven distribution of technological advancement.

      April 1, 2024

      Questions to consider when using AI for PDF data extraction

      Discussing considerations that arise when attempting to automate the extraction of structured data from PDFs and similar documents.

      March 11, 2024

      Two types of startup data problems

      Classifying startups as ML-centric or non-ML is a helpful exercise to uncover the data challenges they’re likely to face.

      March 4, 2024

      Avoiding AI complexity: First, write no code

      Two stories of getting AI functionality to production, which demonstrate the risks inherent in custom development versus starting with a no-code approach.

      February 26, 2024

      Building your startup's minimum viable data stack

      First post in a series on building a minimum viable data stack for startups, introducing key definitions, components, and considerations.

      February 19, 2024

      Nudging ChatGPT to invent books you have no time to read

      Getting ChatGPT Plus to elaborate on possible book content and produce a PDF cheatsheet, with the goal of learning about its capabilities.

      February 12, 2024

      Substance over titles: Your first data hire may be a data scientist

      Advice for hiring a startup’s first data person: match skills to business needs, consider contractors, and get help from data people.

      February 5, 2024

      New decade, new tagline: Data & AI for Impact

      Shifting focus to ‘Data & AI for Impact’, with more startup-related content, increased posting frequency, and deeper audience engagement.

      January 19, 2024

      Supporting volunteer monitoring of marine biodiversity with modern web and data tools

      Summarising the work Uri Seroussi and I did to improve Reef Life Survey’s Reef Species of the World app.

      November 29, 2023

      Lessons from reluctant data engineering

      Video and summary of a talk I gave at DataEngBytes Brisbane on what I learned from doing data engineering as part of every data science role I had.

      October 25, 2023

      My rediscovery of quiet writing on the open web

      Reflections on publishing on this website: Writing publicly to share thoughts and documentation beats chasing views and likes.

      August 28, 2023

      Was data science a failure mode of software engineering?

      Yes, data science projects have suffered from classic software engineering mistakes, but the field is maturing with the rise of new engineering roles.

      June 30, 2023

      How hackable are automated coding assessments?

      Exploring the hackability of speed-based coding tests, using CodeSignal’s Industry Coding Framework as a case study.

      May 26, 2023

      Remaining relevant as a small language model

      Bing Chat recently quipped that humans are small language models. Here are some of my thoughts on how we small language models can remain relevant (for now).

      April 21, 2023

      ChatGPT is transformative AI

      My perspective after a week of using ChatGPT: This is a step change in finding distilled information, and it’s only the beginning.

      December 11, 2022

      Causal Machine Learning is off to a good start, despite some issues

      Reviewing the first three chapters of the book Causal Machine Learning by Robert Osazuwa Ness.

      September 12, 2022

      The mission matters: Moving to climate tech as a data scientist

      Discussing my recent career move into climate tech as a way of doing more to help mitigate dangerous climate change.

      June 6, 2022

      Building useful machine learning tools keeps getting easier: A fish ID case study

      Lessons learned building a fish ID web app with fast.ai and Streamlit, in an attempt to reduce my fear of missing out on the latest deep learning developments.

      March 20, 2022

      Analysis strategies in online A/B experiments: Intention-to-treat, per-protocol, and other lessons from clinical trials

      Epidemiologists analyse clinical trials to estimate the intention-to-treat and per-protocol effects. This post applies their strategies to online experiments.

      January 14, 2022

      Use your human brain to avoid artificial intelligence disasters

      Overview of a talk I gave at a deep learning course, focusing on AI ethics as the need for humans to think on the context and consequences of applying AI.

      November 22, 2021

      Migrating from WordPress.com to Hugo on GitHub + Cloudflare

      My reasons for switching from WordPress.com to Hugo on GitHub + Cloudflare, along with a summary of the solution components and migration process.

      November 10, 2021

      My work with Automattic

      Back-dated meta-post that gathers my posts on Automattic blogs into a summary of the work I’ve done with the company.

      October 7, 2021

      Some highlights from 2020

      Sharing remote teamwork insights, my climate & sustainability activism, Reef Life Survey publications, and progress on Automattic’s Experimentation Platform.

      April 5, 2021

      Many is not enough: Counting simulations to bootstrap the right way

      Going deeper into correct testing of different methods for bootstrap estimation of confidence intervals.

      August 24, 2020

      Software commodities are eating interesting data science work

      Being a data scientist can sometimes feel like a race against software commodities that replace interesting work. What can one do to remain relevant?

      January 11, 2020

      A day in the life of a remote data scientist

      Video of a talk I gave on remote data science work at the Data Science Sydney meetup.

      December 11, 2019

      Bootstrapping the right way?

      Video and summary of a talk I gave at YOW! Data on bootstrap estimation of confidence intervals.

      October 6, 2019

      Hackers beware: Bootstrap sampling may be harmful

      Bootstrap sampling has been promoted as an easy way of modelling uncertainty to hackers without much statistical knowledge. But things aren’t that simple.

      January 7, 2019

      The most practical causal inference book I’ve read (is still a draft)

      Causal Inference by Miguel Hernán and Jamie Robins is a must-read for anyone interested in the area.

      December 24, 2018

      Reflections on remote data science work

      Discussing the pluses and minuses of remote work eighteen months after joining Automattic as a data scientist.

      November 3, 2018

      Defining data science in 2018

      Updating my definition of data science to match changes in the field. It is now broader than before, but its ultimate goal is still to support decisions.

      July 22, 2018

      Advice for aspiring data scientists and other FAQs

      Frequently asked questions by visitors to this site, especially around entering the data science field.

      October 15, 2017

      State of Bandcamp Recommender, Late 2017

      Call for BCRecommender maintainers followed by a decision to shut it down, as I don’t have enough time and Bandcamp now offers recommendations.

      September 2, 2017

      My 10-step path to becoming a remote data scientist with Automattic

      I wanted a well-paid data science-y remote job with an established company that offers a good life balance and makes products I care about. I got it eventually.

      July 29, 2017

      Exploring and visualising Reef Life Survey data

      Web tools I built to visualise Reef Life Survey data and assist citizen scientists in underwater visual census work.

      June 3, 2017

      Customer lifetime value and the proliferation of misinformation on the internet

      There’s a lot of misleading content on the estimation of customer lifetime value. Here’s what I learned about doing it well.

      January 8, 2017

      Ask Why! Finding motives, causes, and purpose in data science

      Video and summary of a talk I gave at the Data Science Sydney meetup, about going beyond the what & how of predictive modelling.

      September 19, 2016

      If you don’t pay attention, data can drive you off a cliff

      Seven common mistakes to avoid when working with data, such as ignoring uncertainty and confusing observed and unobserved quantities.

      August 21, 2016

      Is Data Scientist a useless job title?

      It seems like anyone who touches data can call themselves a data scientist, which makes the title useless. The work they do can still be useful, though.

      August 4, 2016

      Making Bayesian A/B testing more accessible

      A web tool I built to interpret A/B test results in a Bayesian way, including prior specification, visualisations, and decision rules.

      June 19, 2016

      Diving deeper into causality: Pearl, Kleinberg, Hill, and untested assumptions

      Discussing the need for untested assumptions and temporality in causal inference. Mostly based on Samantha Kleinberg’s Causality, Probability, and Time.

      May 14, 2016

      The rise of greedy robots

      Is artificial/machine intelligence a future threat? I argue that it’s already here, with greedy robots already dominating our lives.

      March 20, 2016

      Why you should stop worrying about deep learning and deepen your understanding of causality instead

      Causality is often overlooked but is of much higher relevance to most data scientists than deep learning.

      February 14, 2016

      The joys of offline data collection

      Insights on data collection and machine learning from spending a month sailing, diving, and counting fish with Reef Life Survey.

      January 24, 2016

      This holiday season, give me real insights

      Some companies present raw data or information as “insights”. This post surveys some examples, and discusses how they can be turned into real insights.

      December 8, 2015

      The hardest parts of data science

      Defining feasible problems and coming up with reasonable ways of measuring solutions is harder than building accurate models or obtaining clean data.

      November 23, 2015

      Migrating a simple web application from MongoDB to Elasticsearch

      Migrating BCRecommender from MongoDB to Elasticsearch made it possible to offer a richer search experience to users at a similar cost, among other benefits.

      November 4, 2015

      Miscommunicating science: Simplistic models, nutritionism, and the art of storytelling

      Nutritionism is a special case of misinterpretation and miscommunication of scientific results – something many data scientists encounter in their work.

      October 19, 2015

      The wonderful world of recommender systems

      Giving an overview of the field and common paradigms, and debunking five common myths about recommender systems.

      October 2, 2015

      You don’t need a data scientist (yet)

      Hiring data scientists prematurely is wasteful and frustrating. Here are some questions to ask before you hire your first data scientist.

      August 24, 2015

      Goodbye, Parse.com

      Migrating my web apps away from Parse.com due to reliability issues. Self-hosting is a better solution.

      July 31, 2015

      Learning about deep learning through album cover classification

      Progress on my album cover classification project, highlighting lessons that would be useful to others who are getting started with deep learning.

      July 6, 2015

      Hopping on the deep learning bandwagon

      To become proficient at solving data science problems, you need to get your hands dirty. Here, I used album cover classification to learn about deep learning.

      June 6, 2015

      First steps in data science: author-aware sentiment analysis

      I became a data scientist by doing a PhD, but the same steps can be followed without a formal education program.

      May 2, 2015

      My divestment from fossil fuels

      Recent choices I’ve made to reduce my exposure to fossil fuels, including practical steps that can be taken by Australians and generally applicable lessons.

      April 24, 2015

      My PhD work

      An overview of my PhD in data science / artificial intelligence. Thesis title: Text Mining and Rating Prediction with Topical User Models.

      March 30, 2015

      The long road to a lifestyle business

      Progress since leaving my last full-time job and setting on an independent path that includes data science consulting and work on my own projects.

      March 22, 2015

      Learning to rank for personalised search (Yandex Search Personalisation – Kaggle Competition Summary – Part 2)

      My team’s solution to the Yandex Search Personalisation competition (finished 9th out of 194 teams).

      February 11, 2015

      Is thinking like a search engine possible? (Yandex search personalisation – Kaggle competition summary – part 1)

      Insights on search personalisation and SEO from participating in a Kaggle competition (finished 9th out of 194 teams).

      January 29, 2015

      Automating Parse.com bulk data imports

      A script for importing data into the Parse backend-as-a-service.

      January 15, 2015

      Stochastic Gradient Boosting: Choosing the Best Number of Iterations

      Exploring an approach to choosing the optimal number of iterations in stochastic gradient boosting, following a bug I found in scikit-learn.

      December 29, 2014

      SEO: Mostly about showing up?

      Increasing SEO traffic to BCRecommender by adding content and opening up more pages for crawling. It turns out that thin content is better than no content.

      December 15, 2014

      Fitting noise: Forecasting the sale price of bulldozers (Kaggle competition summary)

      Summary of a Kaggle competition to forecast bulldozer sale price, where I finished 9th out of 476 teams.

      November 19, 2014

      BCRecommender Traction Update

      Update on BCRecommender traction using three channels: blogger outreach, search engine optimisation, and content marketing.

      November 5, 2014

      What is data science?

      Data science has been a hot term in the past few years. Still, there isn’t a single definition of the field. This post discusses my favourite definition.

      October 23, 2014

      Greek Media Monitoring Kaggle competition: My approach

      Summary of my approach to the Greek Media Monitoring Kaggle competition, where I finished 6th out of 120 teams.

      October 7, 2014

      Applying the Traction Book’s Bullseye framework to BCRecommender

      Ranking 19 channels with the goal of getting traction for BCRecommender.

      September 24, 2014

      Bandcamp recommendation and discovery algorithms

      The recommendation backend for my BCRecommender service for personalised Bandcamp music discovery.

      September 19, 2014

      Building a recommender system on a shoestring budget (or: BCRecommender part 2 – general system layout)

      Iterating on my BCRecommender service with the goal of keeping costs low while providing a valuable music recommendation service.

      September 7, 2014

      Building a Bandcamp recommender system (part 1 – motivation)

      My motivation behind building BCRecommender, a free recommendation & discovery service for Bandcamp music.

      August 30, 2014

      How to (almost) win Kaggle competitions

      Summary of a talk I gave at the Data Science Sydney meetup with ten tips on almost-winning Kaggle competitions.

      August 24, 2014

      Data’s hierarchy of needs

      Discussing the hierarchy of needs proposed by Jay Kreps. Key takeaway: Data-driven algorithms & insights can only be as good as the underlying data.

      August 17, 2014

      Kaggle competition tips and summaries

      Pointers to all my Kaggle advice posts and competition summaries.

      April 5, 2014

      Kaggle beginner tips

      First post! An email I sent to members of the Data Science Sydney Meetup with tips on how to get started with Kaggle competitions.

      January 19, 2014
      Text and figures licensed under CC BY-NC-ND 4.0 by Yanir Seroussi, except where noted otherwise  | Powered by Hugo & PaperMod
      \ No newline at end of file diff --git a/posts/index.xml b/posts/index.xml index 352eb885c..2a3f71886 100644 --- a/posts/index.xml +++ b/posts/index.xml @@ -1 +1 @@ -Browse Posts on Yanir Seroussi | Data & AI for Startup Impacthttps://yanirseroussi.com/posts/Recent content in Browse Posts on Yanir Seroussi | Data & AI for Startup ImpactHugo -- gohugo.ioen-auText and figures licensed under [CC BY-NC-ND 4.0](https://creativecommons.org/licenses/by-nc-nd/4.0/) by [Yanir Seroussi](https://yanirseroussi.com/about/), except where noted otherwiseHow to avoid startups with poor development processeshttps://yanirseroussi.com/2024/06/03/how-to-avoid-startups-with-poor-development-processes/Mon, 03 Jun 2024 02:45:00 +0000https://yanirseroussi.com/2024/06/03/how-to-avoid-startups-with-poor-development-processes/Questions that prospective data specialists and engineers should ask about development processes before accepting a startup role.Plumbing, Decisions, and Automation: De-hyping Data & AIhttps://yanirseroussi.com/2024/05/27/plumbing-decisions-and-automation-de-hyping-data-and-ai/Mon, 27 May 2024 02:00:00 +0000https://yanirseroussi.com/2024/05/27/plumbing-decisions-and-automation-de-hyping-data-and-ai/Three essential questions to understand where an organisation stands when it comes to Data &amp; AI (with zero hype).Question startup culture before accepting a data-to-AI rolehttps://yanirseroussi.com/2024/05/20/question-startup-culture-before-accepting-a-data-to-ai-role/Mon, 20 May 2024 02:25:00 +0000https://yanirseroussi.com/2024/05/20/question-startup-culture-before-accepting-a-data-to-ai-role/Eight questions that prospective data-to-AI employees should ask about a startup&rsquo;s work and data culture.Probing the People aspects of an early-stage startuphttps://yanirseroussi.com/2024/05/13/probing-the-people-aspects-of-an-early-stage-startup/Mon, 13 May 2024 02:00:00 +0000https://yanirseroussi.com/2024/05/13/probing-the-people-aspects-of-an-early-stage-startup/Ten questions that prospective employees should ask about a startup&rsquo;s team, especially for data-centric roles.Business questions to ask before taking a startup data rolehttps://yanirseroussi.com/2024/05/06/business-questions-to-ask-before-taking-a-startup-data-role/Mon, 06 May 2024 04:30:00 +0000https://yanirseroussi.com/2024/05/06/business-questions-to-ask-before-taking-a-startup-data-role/Fourteen questions that prospective employees should ask about a startup&rsquo;s business model and product, especially for data-focused roles.Mentorship and the art of actionable advicehttps://yanirseroussi.com/2024/04/29/mentorship-and-the-art-of-actionable-advice/Mon, 29 Apr 2024 06:30:00 +0000https://yanirseroussi.com/2024/04/29/mentorship-and-the-art-of-actionable-advice/Reflections on what it takes to package expertise and deliver timely, actionable advice outside the context of employee relationships.Assessing a startup's data-to-AI healthhttps://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/Mon, 22 Apr 2024 06:00:00 +0000https://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/Reviewing the areas that should be assessed to determine a startup&rsquo;s opportunities and challenges on the data/AI/ML front.AI does not obviate the need for testing and observabilityhttps://yanirseroussi.com/2024/04/15/ai-does-not-obviate-the-need-for-testing-and-observability/Mon, 15 Apr 2024 05:00:00 +0000https://yanirseroussi.com/2024/04/15/ai-does-not-obviate-the-need-for-testing-and-observability/It&rsquo;s easy to prototype with AI, but production-grade AI apps require even more thorough testing and observability than traditional software.My experience as a Data Tech Lead with Work on Climatehttps://yanirseroussi.com/2024/04/08/my-experience-as-a-data-tech-lead-with-work-on-climate/Mon, 08 Apr 2024 02:00:00 +0000https://yanirseroussi.com/2024/04/08/my-experience-as-a-data-tech-lead-with-work-on-climate/The story of how I joined Work on Climate as a volunteer and became its data tech lead, with lessons applied to consulting &amp; fractional work.Artificial intelligence, automation, and the art of counting fishhttps://yanirseroussi.com/2024/04/01/artificial-intelligence-automation-and-the-art-of-counting-fish/Mon, 01 Apr 2024 06:00:00 +0000https://yanirseroussi.com/2024/04/01/artificial-intelligence-automation-and-the-art-of-counting-fish/Discussing the use of AI to automate underwater marine surveys as an example of the uneven distribution of technological advancement.Questions to consider when using AI for PDF data extractionhttps://yanirseroussi.com/2024/03/11/questions-to-consider-when-using-ai-for-pdf-data-extraction/Mon, 11 Mar 2024 00:00:00 +0000https://yanirseroussi.com/2024/03/11/questions-to-consider-when-using-ai-for-pdf-data-extraction/Discussing considerations that arise when attempting to automate the extraction of structured data from PDFs and similar documents.Two types of startup data problemshttps://yanirseroussi.com/2024/03/04/two-types-of-startup-data-problems/Mon, 04 Mar 2024 02:00:00 +0000https://yanirseroussi.com/2024/03/04/two-types-of-startup-data-problems/Classifying startups as ML-centric or non-ML is a helpful exercise to uncover the data challenges they&rsquo;re likely to face.Avoiding AI complexity: First, write no codehttps://yanirseroussi.com/2024/02/26/avoiding-ai-complexity-first-write-no-code/Mon, 26 Feb 2024 01:45:00 +0000https://yanirseroussi.com/2024/02/26/avoiding-ai-complexity-first-write-no-code/Two stories of getting AI functionality to production, which demonstrate the risks inherent in custom development versus starting with a no-code approach.Building your startup's minimum viable data stackhttps://yanirseroussi.com/2024/02/19/building-your-startups-minimum-viable-data-stack/Mon, 19 Feb 2024 00:00:00 +0000https://yanirseroussi.com/2024/02/19/building-your-startups-minimum-viable-data-stack/First post in a series on building a minimum viable data stack for startups, introducing key definitions, components, and considerations.Nudging ChatGPT to invent books you have no time to readhttps://yanirseroussi.com/2024/02/12/nudging-chatgpt-to-invent-books-you-have-no-time-to-read/Mon, 12 Feb 2024 05:00:00 +0000https://yanirseroussi.com/2024/02/12/nudging-chatgpt-to-invent-books-you-have-no-time-to-read/Getting ChatGPT Plus to elaborate on possible book content and produce a PDF cheatsheet, with the goal of learning about its capabilities.Substance over titles: Your first data hire may be a data scientisthttps://yanirseroussi.com/2024/02/05/substance-over-titles-your-first-data-hire-may-be-a-data-scientist/Mon, 05 Feb 2024 02:45:00 +0000https://yanirseroussi.com/2024/02/05/substance-over-titles-your-first-data-hire-may-be-a-data-scientist/Advice for hiring a startup&rsquo;s first data person: match skills to business needs, consider contractors, and get help from data people.New decade, new tagline: Data & AI for Impacthttps://yanirseroussi.com/2024/01/19/new-decade-new-tagline-data-and-ai-for-impact/Fri, 19 Jan 2024 00:00:00 +0000https://yanirseroussi.com/2024/01/19/new-decade-new-tagline-data-and-ai-for-impact/Shifting focus to &lsquo;Data &amp; AI for Impact&rsquo;, with more startup-related content, increased posting frequency, and deeper audience engagement.Supporting volunteer monitoring of marine biodiversity with modern web and data toolshttps://yanirseroussi.com/2023/11/29/supporting-volunteer-monitoring-of-marine-biodiversity-with-modern-web-and-data-tools/Wed, 29 Nov 2023 02:00:00 +0000https://yanirseroussi.com/2023/11/29/supporting-volunteer-monitoring-of-marine-biodiversity-with-modern-web-and-data-tools/Summarising the work Uri Seroussi and I did to improve Reef Life Survey&rsquo;s Reef Species of the World app.Lessons from reluctant data engineeringhttps://yanirseroussi.com/2023/10/25/lessons-from-reluctant-data-engineering/Wed, 25 Oct 2023 04:45:00 +0000https://yanirseroussi.com/2023/10/25/lessons-from-reluctant-data-engineering/Video and summary of a talk I gave at DataEngBytes Brisbane on what I learned from doing data engineering as part of every data science role I had.My rediscovery of quiet writing on the open webhttps://yanirseroussi.com/2023/08/28/my-rediscovery-of-quiet-writing-on-the-open-web/Mon, 28 Aug 2023 05:30:00 +0000https://yanirseroussi.com/2023/08/28/my-rediscovery-of-quiet-writing-on-the-open-web/Reflections on publishing on this website: Writing publicly to share thoughts and documentation beats chasing views and likes.Was data science a failure mode of software engineering?https://yanirseroussi.com/2023/06/30/was-data-science-a-failure-mode-of-software-engineering/Fri, 30 Jun 2023 00:06:30 +0000https://yanirseroussi.com/2023/06/30/was-data-science-a-failure-mode-of-software-engineering/Yes, data science projects have suffered from classic software engineering mistakes, but the field is maturing with the rise of new engineering roles.How hackable are automated coding assessments?https://yanirseroussi.com/2023/05/26/how-hackable-are-automated-coding-assessments/Fri, 26 May 2023 00:03:00 +0000https://yanirseroussi.com/2023/05/26/how-hackable-are-automated-coding-assessments/Exploring the hackability of speed-based coding tests, using CodeSignal&rsquo;s Industry Coding Framework as a case study.Remaining relevant as a small language modelhttps://yanirseroussi.com/2023/04/21/remaining-relevant-as-a-small-language-model/Fri, 21 Apr 2023 00:06:30 +0000https://yanirseroussi.com/2023/04/21/remaining-relevant-as-a-small-language-model/Bing Chat recently quipped that humans are small language models. Here are some of my thoughts on how we small language models can remain relevant (for now).ChatGPT is transformative AIhttps://yanirseroussi.com/2022/12/11/chatgpt-is-transformative-ai/Sun, 11 Dec 2022 00:00:00 +0000https://yanirseroussi.com/2022/12/11/chatgpt-is-transformative-ai/My perspective after a week of using ChatGPT: This is a step change in finding distilled information, and it&rsquo;s only the beginning.Causal Machine Learning is off to a good start, despite some issueshttps://yanirseroussi.com/2022/09/12/causal-machine-learning-book-draft-review/Mon, 12 Sep 2022 02:45:00 +0000https://yanirseroussi.com/2022/09/12/causal-machine-learning-book-draft-review/Reviewing the first three chapters of the book Causal Machine Learning by Robert Osazuwa Ness.The mission matters: Moving to climate tech as a data scientisthttps://yanirseroussi.com/2022/06/06/the-mission-matters-moving-to-climate-tech-as-a-data-scientist/Mon, 06 Jun 2022 00:00:00 +0000https://yanirseroussi.com/2022/06/06/the-mission-matters-moving-to-climate-tech-as-a-data-scientist/Discussing my recent career move into climate tech as a way of doing more to help mitigate dangerous climate change.Building useful machine learning tools keeps getting easier: A fish ID case studyhttps://yanirseroussi.com/2022/03/20/building-useful-machine-learning-tools-keeps-getting-easier-a-fish-id-case-study/Sun, 20 Mar 2022 04:30:00 +0000https://yanirseroussi.com/2022/03/20/building-useful-machine-learning-tools-keeps-getting-easier-a-fish-id-case-study/Lessons learned building a fish ID web app with fast.ai and Streamlit, in an attempt to reduce my fear of missing out on the latest deep learning developments.Analysis strategies in online A/B experiments: Intention-to-treat, per-protocol, and other lessons from clinical trialshttps://yanirseroussi.com/2022/01/14/analysis-strategies-in-online-a-b-experiments/Fri, 14 Jan 2022 00:05:40 +0000https://yanirseroussi.com/2022/01/14/analysis-strategies-in-online-a-b-experiments/Epidemiologists analyse clinical trials to estimate the intention-to-treat and per-protocol effects. This post applies their strategies to online experiments.Use your human brain to avoid artificial intelligence disastershttps://yanirseroussi.com/2021/11/22/use-your-human-brain-to-avoid-artificial-intelligence-disasters/Mon, 22 Nov 2021 03:45:00 +0000https://yanirseroussi.com/2021/11/22/use-your-human-brain-to-avoid-artificial-intelligence-disasters/Overview of a talk I gave at a deep learning course, focusing on AI ethics as the need for humans to think on the context and consequences of applying AI.Migrating from WordPress.com to Hugo on GitHub + Cloudflarehttps://yanirseroussi.com/2021/11/10/migrating-from-wordpress-com-to-hugo-on-github-cloudflare/Wed, 10 Nov 2021 06:30:00 +0000https://yanirseroussi.com/2021/11/10/migrating-from-wordpress-com-to-hugo-on-github-cloudflare/My reasons for switching from WordPress.com to Hugo on GitHub + Cloudflare, along with a summary of the solution components and migration process.My work with Automattichttps://yanirseroussi.com/2021/10/07/my-work-with-automattic/Thu, 07 Oct 2021 00:00:00 +0000https://yanirseroussi.com/2021/10/07/my-work-with-automattic/Back-dated meta-post that gathers my posts on Automattic blogs into a summary of the work I&rsquo;ve done with the company.Some highlights from 2020https://yanirseroussi.com/2021/04/05/some-highlights-from-2020/Mon, 05 Apr 2021 06:41:48 +0000https://yanirseroussi.com/2021/04/05/some-highlights-from-2020/Sharing remote teamwork insights, my climate &amp; sustainability activism, Reef Life Survey publications, and progress on Automattic&rsquo;s Experimentation Platform.Many is not enough: Counting simulations to bootstrap the right wayhttps://yanirseroussi.com/2020/08/24/many-is-not-enough-counting-simulations-to-bootstrap-the-right-way/Mon, 24 Aug 2020 01:35:17 +0000https://yanirseroussi.com/2020/08/24/many-is-not-enough-counting-simulations-to-bootstrap-the-right-way/Going deeper into correct testing of different methods for bootstrap estimation of confidence intervals.Software commodities are eating interesting data science workhttps://yanirseroussi.com/2020/01/11/software-commodities-are-eating-interesting-data-science-work/Sat, 11 Jan 2020 09:22:35 +0000https://yanirseroussi.com/2020/01/11/software-commodities-are-eating-interesting-data-science-work/Being a data scientist can sometimes feel like a race against software commodities that replace interesting work. What can one do to remain relevant?A day in the life of a remote data scientisthttps://yanirseroussi.com/2019/12/12/a-day-in-the-life-of-a-remote-data-scientist/Wed, 11 Dec 2019 22:06:19 +0000https://yanirseroussi.com/2019/12/12/a-day-in-the-life-of-a-remote-data-scientist/Video of a talk I gave on remote data science work at the Data Science Sydney meetup.Bootstrapping the right way?https://yanirseroussi.com/2019/10/06/bootstrapping-the-right-way/Sun, 06 Oct 2019 06:48:07 +0000https://yanirseroussi.com/2019/10/06/bootstrapping-the-right-way/Video and summary of a talk I gave at YOW! Data on bootstrap estimation of confidence intervals.Hackers beware: Bootstrap sampling may be harmfulhttps://yanirseroussi.com/2019/01/08/hackers-beware-bootstrap-sampling-may-be-harmful/Mon, 07 Jan 2019 21:07:56 +0000https://yanirseroussi.com/2019/01/08/hackers-beware-bootstrap-sampling-may-be-harmful/Bootstrap sampling has been promoted as an easy way of modelling uncertainty to hackers without much statistical knowledge. But things aren&rsquo;t that simple.The most practical causal inference book I’ve read (is still a draft)https://yanirseroussi.com/2018/12/24/the-most-practical-causal-inference-book-ive-read-is-still-a-draft/Mon, 24 Dec 2018 02:37:50 +0000https://yanirseroussi.com/2018/12/24/the-most-practical-causal-inference-book-ive-read-is-still-a-draft/Causal Inference by Miguel Hernán and Jamie Robins is a must-read for anyone interested in the area.Reflections on remote data science workhttps://yanirseroussi.com/2018/11/03/reflections-on-remote-data-science-work/Sat, 03 Nov 2018 06:33:13 +0000https://yanirseroussi.com/2018/11/03/reflections-on-remote-data-science-work/Discussing the pluses and minuses of remote work eighteen months after joining Automattic as a data scientist.Defining data science in 2018https://yanirseroussi.com/2018/07/22/defining-data-science-in-2018/Sun, 22 Jul 2018 08:27:43 +0000https://yanirseroussi.com/2018/07/22/defining-data-science-in-2018/Updating my definition of data science to match changes in the field. It is now broader than before, but its ultimate goal is still to support decisions.Advice for aspiring data scientists and other FAQshttps://yanirseroussi.com/2017/10/15/advice-for-aspiring-data-scientists-and-other-faqs/Sun, 15 Oct 2017 09:15:25 +0000https://yanirseroussi.com/2017/10/15/advice-for-aspiring-data-scientists-and-other-faqs/Frequently asked questions by visitors to this site, especially around entering the data science field.State of Bandcamp Recommender, Late 2017https://yanirseroussi.com/2017/09/02/state-of-bandcamp-recommender/Sat, 02 Sep 2017 10:19:02 +0000https://yanirseroussi.com/2017/09/02/state-of-bandcamp-recommender/Call for BCRecommender maintainers followed by a decision to shut it down, as I don&rsquo;t have enough time and Bandcamp now offers recommendations.My 10-step path to becoming a remote data scientist with Automattichttps://yanirseroussi.com/2017/07/29/my-10-step-path-to-becoming-a-remote-data-scientist-with-automattic/Sat, 29 Jul 2017 05:39:26 +0000https://yanirseroussi.com/2017/07/29/my-10-step-path-to-becoming-a-remote-data-scientist-with-automattic/I wanted a well-paid data science-y remote job with an established company that offers a good life balance and makes products I care about. I got it eventually.Exploring and visualising Reef Life Survey datahttps://yanirseroussi.com/2017/06/03/exploring-and-visualising-reef-life-survey-data/Sat, 03 Jun 2017 00:49:05 +0000https://yanirseroussi.com/2017/06/03/exploring-and-visualising-reef-life-survey-data/Web tools I built to visualise Reef Life Survey data and assist citizen scientists in underwater visual census work.Customer lifetime value and the proliferation of misinformation on the internethttps://yanirseroussi.com/2017/01/08/customer-lifetime-value-and-the-proliferation-of-misinformation-on-the-internet/Sun, 08 Jan 2017 20:02:30 +0000https://yanirseroussi.com/2017/01/08/customer-lifetime-value-and-the-proliferation-of-misinformation-on-the-internet/There&rsquo;s a lot of misleading content on the estimation of customer lifetime value. Here&rsquo;s what I learned about doing it well.Ask Why! Finding motives, causes, and purpose in data sciencehttps://yanirseroussi.com/2016/09/19/ask-why-finding-motives-causes-and-purpose-in-data-science/Mon, 19 Sep 2016 21:28:44 +0000https://yanirseroussi.com/2016/09/19/ask-why-finding-motives-causes-and-purpose-in-data-science/Video and summary of a talk I gave at the Data Science Sydney meetup, about going beyond the what &amp; how of predictive modelling.If you don’t pay attention, data can drive you off a cliffhttps://yanirseroussi.com/2016/08/21/seven-ways-to-be-data-driven-off-a-cliff/Sun, 21 Aug 2016 21:34:17 +0000https://yanirseroussi.com/2016/08/21/seven-ways-to-be-data-driven-off-a-cliff/Seven common mistakes to avoid when working with data, such as ignoring uncertainty and confusing observed and unobserved quantities.Is Data Scientist a useless job title?https://yanirseroussi.com/2016/08/04/is-data-scientist-a-useless-job-title/Thu, 04 Aug 2016 22:26:03 +0000https://yanirseroussi.com/2016/08/04/is-data-scientist-a-useless-job-title/It seems like anyone who touches data can call themselves a data scientist, which makes the title useless. The work they do can still be useful, though.Making Bayesian A/B testing more accessiblehttps://yanirseroussi.com/2016/06/19/making-bayesian-ab-testing-more-accessible/Sun, 19 Jun 2016 10:32:15 +0000https://yanirseroussi.com/2016/06/19/making-bayesian-ab-testing-more-accessible/A web tool I built to interpret A/B test results in a Bayesian way, including prior specification, visualisations, and decision rules.Diving deeper into causality: Pearl, Kleinberg, Hill, and untested assumptionshttps://yanirseroussi.com/2016/05/15/diving-deeper-into-causality-pearl-kleinberg-hill-and-untested-assumptions/Sat, 14 May 2016 19:57:03 +0000https://yanirseroussi.com/2016/05/15/diving-deeper-into-causality-pearl-kleinberg-hill-and-untested-assumptions/Discussing the need for untested assumptions and temporality in causal inference. Mostly based on Samantha Kleinberg&rsquo;s Causality, Probability, and Time.The rise of greedy robotshttps://yanirseroussi.com/2016/03/20/the-rise-of-greedy-robots/Sun, 20 Mar 2016 20:33:43 +0000https://yanirseroussi.com/2016/03/20/the-rise-of-greedy-robots/Is artificial/machine intelligence a future threat? I argue that it&rsquo;s already here, with greedy robots already dominating our lives.Why you should stop worrying about deep learning and deepen your understanding of causality insteadhttps://yanirseroussi.com/2016/02/14/why-you-should-stop-worrying-about-deep-learning-and-deepen-your-understanding-of-causality-instead/Sun, 14 Feb 2016 11:04:11 +0000https://yanirseroussi.com/2016/02/14/why-you-should-stop-worrying-about-deep-learning-and-deepen-your-understanding-of-causality-instead/Causality is often overlooked but is of much higher relevance to most data scientists than deep learning.The joys of offline data collectionhttps://yanirseroussi.com/2016/01/24/the-joys-of-offline-data-collection/Sun, 24 Jan 2016 00:32:25 +0000https://yanirseroussi.com/2016/01/24/the-joys-of-offline-data-collection/Insights on data collection and machine learning from spending a month sailing, diving, and counting fish with Reef Life Survey.This holiday season, give me real insightshttps://yanirseroussi.com/2015/12/08/this-holiday-season-give-me-real-insights/Tue, 08 Dec 2015 06:57:25 +0000https://yanirseroussi.com/2015/12/08/this-holiday-season-give-me-real-insights/Some companies present raw data or information as &ldquo;insights&rdquo;. This post surveys some examples, and discusses how they can be turned into real insights.The hardest parts of data sciencehttps://yanirseroussi.com/2015/11/23/the-hardest-parts-of-data-science/Mon, 23 Nov 2015 04:14:21 +0000https://yanirseroussi.com/2015/11/23/the-hardest-parts-of-data-science/Defining feasible problems and coming up with reasonable ways of measuring solutions is harder than building accurate models or obtaining clean data.Migrating a simple web application from MongoDB to Elasticsearchhttps://yanirseroussi.com/2015/11/04/migrating-a-simple-web-application-from-mongodb-to-elasticsearch/Wed, 04 Nov 2015 03:53:18 +0000https://yanirseroussi.com/2015/11/04/migrating-a-simple-web-application-from-mongodb-to-elasticsearch/Migrating BCRecommender from MongoDB to Elasticsearch made it possible to offer a richer search experience to users at a similar cost, among other benefits.Miscommunicating science: Simplistic models, nutritionism, and the art of storytellinghttps://yanirseroussi.com/2015/10/19/nutritionism-and-the-need-for-complex-models-to-explain-complex-phenomena/Mon, 19 Oct 2015 00:02:32 +0000https://yanirseroussi.com/2015/10/19/nutritionism-and-the-need-for-complex-models-to-explain-complex-phenomena/Nutritionism is a special case of misinterpretation and miscommunication of scientific results – something many data scientists encounter in their work.The wonderful world of recommender systemshttps://yanirseroussi.com/2015/10/02/the-wonderful-world-of-recommender-systems/Fri, 02 Oct 2015 05:25:57 +0000https://yanirseroussi.com/2015/10/02/the-wonderful-world-of-recommender-systems/Giving an overview of the field and common paradigms, and debunking five common myths about recommender systems.You don’t need a data scientist (yet)https://yanirseroussi.com/2015/08/24/you-dont-need-a-data-scientist-yet/Mon, 24 Aug 2015 08:25:30 +0000https://yanirseroussi.com/2015/08/24/you-dont-need-a-data-scientist-yet/Hiring data scientists prematurely is wasteful and frustrating. Here are some questions to ask before you hire your first data scientist.Goodbye, Parse.comhttps://yanirseroussi.com/2015/07/31/goodbye-parse-com/Fri, 31 Jul 2015 03:29:50 +0000https://yanirseroussi.com/2015/07/31/goodbye-parse-com/Migrating my web apps away from Parse.com due to reliability issues. Self-hosting is a better solution.Learning about deep learning through album cover classificationhttps://yanirseroussi.com/2015/07/06/learning-about-deep-learning-through-album-cover-classification/Mon, 06 Jul 2015 22:21:42 +0000https://yanirseroussi.com/2015/07/06/learning-about-deep-learning-through-album-cover-classification/Progress on my album cover classification project, highlighting lessons that would be useful to others who are getting started with deep learning.Hopping on the deep learning bandwagonhttps://yanirseroussi.com/2015/06/06/hopping-on-the-deep-learning-bandwagon/Sat, 06 Jun 2015 05:00:22 +0000https://yanirseroussi.com/2015/06/06/hopping-on-the-deep-learning-bandwagon/To become proficient at solving data science problems, you need to get your hands dirty. Here, I used album cover classification to learn about deep learning.First steps in data science: author-aware sentiment analysishttps://yanirseroussi.com/2015/05/02/first-steps-in-data-science-author-aware-sentiment-analysis/Sat, 02 May 2015 08:31:10 +0000https://yanirseroussi.com/2015/05/02/first-steps-in-data-science-author-aware-sentiment-analysis/I became a data scientist by doing a PhD, but the same steps can be followed without a formal education program.My divestment from fossil fuelshttps://yanirseroussi.com/2015/04/24/my-divestment-from-fossil-fuels/Fri, 24 Apr 2015 00:19:36 +0000https://yanirseroussi.com/2015/04/24/my-divestment-from-fossil-fuels/Recent choices I&rsquo;ve made to reduce my exposure to fossil fuels, including practical steps that can be taken by Australians and generally applicable lessons.My PhD workhttps://yanirseroussi.com/phd-work/Mon, 30 Mar 2015 03:23:33 +0000https://yanirseroussi.com/phd-work/An overview of my PhD in data science / artificial intelligence. Thesis title: Text Mining and Rating Prediction with Topical User Models.The long road to a lifestyle businesshttps://yanirseroussi.com/2015/03/22/the-long-road-to-a-lifestyle-business/Sun, 22 Mar 2015 09:43:47 +0000https://yanirseroussi.com/2015/03/22/the-long-road-to-a-lifestyle-business/Progress since leaving my last full-time job and setting on an independent path that includes data science consulting and work on my own projects.Learning to rank for personalised search (Yandex Search Personalisation – Kaggle Competition Summary – Part 2)https://yanirseroussi.com/2015/02/11/learning-to-rank-for-personalised-search-yandex-search-personalisation-kaggle-competition-summary-part-2/Wed, 11 Feb 2015 06:34:17 +0000https://yanirseroussi.com/2015/02/11/learning-to-rank-for-personalised-search-yandex-search-personalisation-kaggle-competition-summary-part-2/My team&rsquo;s solution to the Yandex Search Personalisation competition (finished 9th out of 194 teams).Is thinking like a search engine possible? (Yandex search personalisation – Kaggle competition summary – part 1)https://yanirseroussi.com/2015/01/29/is-thinking-like-a-search-engine-possible-yandex-search-personalisation-kaggle-competition-summary-part-1/Thu, 29 Jan 2015 10:37:39 +0000https://yanirseroussi.com/2015/01/29/is-thinking-like-a-search-engine-possible-yandex-search-personalisation-kaggle-competition-summary-part-1/Insights on search personalisation and SEO from participating in a Kaggle competition (finished 9th out of 194 teams).Automating Parse.com bulk data importshttps://yanirseroussi.com/2015/01/15/automating-parse-com-bulk-data-imports/Thu, 15 Jan 2015 04:41:16 +0000https://yanirseroussi.com/2015/01/15/automating-parse-com-bulk-data-imports/A script for importing data into the Parse backend-as-a-service.Stochastic Gradient Boosting: Choosing the Best Number of Iterationshttps://yanirseroussi.com/2014/12/29/stochastic-gradient-boosting-choosing-the-best-number-of-iterations/Mon, 29 Dec 2014 02:30:06 +0000https://yanirseroussi.com/2014/12/29/stochastic-gradient-boosting-choosing-the-best-number-of-iterations/Exploring an approach to choosing the optimal number of iterations in stochastic gradient boosting, following a bug I found in scikit-learn.SEO: Mostly about showing up?https://yanirseroussi.com/2014/12/15/seo-mostly-about-showing-up/Mon, 15 Dec 2014 04:25:25 +0000https://yanirseroussi.com/2014/12/15/seo-mostly-about-showing-up/Increasing SEO traffic to BCRecommender by adding content and opening up more pages for crawling. It turns out that thin content is better than no content.Fitting noise: Forecasting the sale price of bulldozers (Kaggle competition summary)https://yanirseroussi.com/2014/11/19/fitting-noise-forecasting-the-sale-price-of-bulldozers-kaggle-competition-summary/Wed, 19 Nov 2014 09:17:34 +0000https://yanirseroussi.com/2014/11/19/fitting-noise-forecasting-the-sale-price-of-bulldozers-kaggle-competition-summary/Summary of a Kaggle competition to forecast bulldozer sale price, where I finished 9th out of 476 teams.BCRecommender Traction Updatehttps://yanirseroussi.com/2014/11/05/bcrecommender-traction-update/Wed, 05 Nov 2014 02:29:35 +0000https://yanirseroussi.com/2014/11/05/bcrecommender-traction-update/Update on BCRecommender traction using three channels: blogger outreach, search engine optimisation, and content marketing.What is data science?https://yanirseroussi.com/2014/10/23/what-is-data-science/Thu, 23 Oct 2014 03:22:08 +0000https://yanirseroussi.com/2014/10/23/what-is-data-science/Data science has been a hot term in the past few years. Still, there isn&rsquo;t a single definition of the field. This post discusses my favourite definition.Greek Media Monitoring Kaggle competition: My approachhttps://yanirseroussi.com/2014/10/07/greek-media-monitoring-kaggle-competition-my-approach/Tue, 07 Oct 2014 03:21:35 +0000https://yanirseroussi.com/2014/10/07/greek-media-monitoring-kaggle-competition-my-approach/Summary of my approach to the Greek Media Monitoring Kaggle competition, where I finished 6th out of 120 teams.Applying the Traction Book’s Bullseye framework to BCRecommenderhttps://yanirseroussi.com/2014/09/24/applying-the-traction-books-bullseye-framework-to-bcrecommender/Wed, 24 Sep 2014 04:57:39 +0000https://yanirseroussi.com/2014/09/24/applying-the-traction-books-bullseye-framework-to-bcrecommender/Ranking 19 channels with the goal of getting traction for BCRecommender.Bandcamp recommendation and discovery algorithmshttps://yanirseroussi.com/2014/09/19/bandcamp-recommendation-and-discovery-algorithms/Fri, 19 Sep 2014 14:26:55 +0000https://yanirseroussi.com/2014/09/19/bandcamp-recommendation-and-discovery-algorithms/The recommendation backend for my BCRecommender service for personalised Bandcamp music discovery.Building a recommender system on a shoestring budget (or: BCRecommender part 2 – general system layout)https://yanirseroussi.com/2014/09/07/building-a-recommender-system-on-a-shoestring-budget/Sun, 07 Sep 2014 10:48:44 +0000https://yanirseroussi.com/2014/09/07/building-a-recommender-system-on-a-shoestring-budget/Iterating on my BCRecommender service with the goal of keeping costs low while providing a valuable music recommendation service.Building a Bandcamp recommender system (part 1 – motivation)https://yanirseroussi.com/2014/08/30/building-a-bandcamp-recommender-system-part-1-motivation/Sat, 30 Aug 2014 08:11:38 +0000https://yanirseroussi.com/2014/08/30/building-a-bandcamp-recommender-system-part-1-motivation/My motivation behind building BCRecommender, a free recommendation &amp; discovery service for Bandcamp music.How to (almost) win Kaggle competitionshttps://yanirseroussi.com/2014/08/24/how-to-almost-win-kaggle-competitions/Sun, 24 Aug 2014 12:40:53 +0000https://yanirseroussi.com/2014/08/24/how-to-almost-win-kaggle-competitions/Summary of a talk I gave at the Data Science Sydney meetup with ten tips on almost-winning Kaggle competitions.Data’s hierarchy of needshttps://yanirseroussi.com/2014/08/17/datas-hierarchy-of-needs/Sun, 17 Aug 2014 13:09:30 +0000https://yanirseroussi.com/2014/08/17/datas-hierarchy-of-needs/Discussing the hierarchy of needs proposed by Jay Kreps. Key takeaway: Data-driven algorithms &amp; insights can only be as good as the underlying data.Kaggle competition tips and summarieshttps://yanirseroussi.com/kaggle/Sat, 05 Apr 2014 23:46:10 +0000https://yanirseroussi.com/kaggle/Pointers to all my Kaggle advice posts and competition summaries.Kaggle beginner tipshttps://yanirseroussi.com/2014/01/19/kaggle-beginner-tips/Sun, 19 Jan 2014 10:34:28 +0000https://yanirseroussi.com/2014/01/19/kaggle-beginner-tips/First post! An email I sent to members of the Data Science Sydney Meetup with tips on how to get started with Kaggle competitions. \ No newline at end of file +Browse Posts on Yanir Seroussi | Data & AI for Startup Impacthttps://yanirseroussi.com/posts/Recent content in Browse Posts on Yanir Seroussi | Data & AI for Startup ImpactHugo -- gohugo.ioen-auText and figures licensed under [CC BY-NC-ND 4.0](https://creativecommons.org/licenses/by-nc-nd/4.0/) by [Yanir Seroussi](https://yanirseroussi.com/about/), except where noted otherwiseStartup data health starts with healthy event trackinghttps://yanirseroussi.com/2024/06/10/startup-data-health-starts-with-healthy-event-tracking/Mon, 10 Jun 2024 04:00:00 +0000https://yanirseroussi.com/2024/06/10/startup-data-health-starts-with-healthy-event-tracking/Expanding on the startup health check question of tracking Kukuyeva&rsquo;s five business aspects as wide events.How to avoid startups with poor development processeshttps://yanirseroussi.com/2024/06/03/how-to-avoid-startups-with-poor-development-processes/Mon, 03 Jun 2024 02:45:00 +0000https://yanirseroussi.com/2024/06/03/how-to-avoid-startups-with-poor-development-processes/Questions that prospective data specialists and engineers should ask about development processes before accepting a startup role.Plumbing, Decisions, and Automation: De-hyping Data & AIhttps://yanirseroussi.com/2024/05/27/plumbing-decisions-and-automation-de-hyping-data-and-ai/Mon, 27 May 2024 02:00:00 +0000https://yanirseroussi.com/2024/05/27/plumbing-decisions-and-automation-de-hyping-data-and-ai/Three essential questions to understand where an organisation stands when it comes to Data &amp; AI (with zero hype).Question startup culture before accepting a data-to-AI rolehttps://yanirseroussi.com/2024/05/20/question-startup-culture-before-accepting-a-data-to-ai-role/Mon, 20 May 2024 02:25:00 +0000https://yanirseroussi.com/2024/05/20/question-startup-culture-before-accepting-a-data-to-ai-role/Eight questions that prospective data-to-AI employees should ask about a startup&rsquo;s work and data culture.Probing the People aspects of an early-stage startuphttps://yanirseroussi.com/2024/05/13/probing-the-people-aspects-of-an-early-stage-startup/Mon, 13 May 2024 02:00:00 +0000https://yanirseroussi.com/2024/05/13/probing-the-people-aspects-of-an-early-stage-startup/Ten questions that prospective employees should ask about a startup&rsquo;s team, especially for data-centric roles.Business questions to ask before taking a startup data rolehttps://yanirseroussi.com/2024/05/06/business-questions-to-ask-before-taking-a-startup-data-role/Mon, 06 May 2024 04:30:00 +0000https://yanirseroussi.com/2024/05/06/business-questions-to-ask-before-taking-a-startup-data-role/Fourteen questions that prospective employees should ask about a startup&rsquo;s business model and product, especially for data-focused roles.Mentorship and the art of actionable advicehttps://yanirseroussi.com/2024/04/29/mentorship-and-the-art-of-actionable-advice/Mon, 29 Apr 2024 06:30:00 +0000https://yanirseroussi.com/2024/04/29/mentorship-and-the-art-of-actionable-advice/Reflections on what it takes to package expertise and deliver timely, actionable advice outside the context of employee relationships.Assessing a startup's data-to-AI healthhttps://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/Mon, 22 Apr 2024 06:00:00 +0000https://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/Reviewing the areas that should be assessed to determine a startup&rsquo;s opportunities and challenges on the data/AI/ML front.AI does not obviate the need for testing and observabilityhttps://yanirseroussi.com/2024/04/15/ai-does-not-obviate-the-need-for-testing-and-observability/Mon, 15 Apr 2024 05:00:00 +0000https://yanirseroussi.com/2024/04/15/ai-does-not-obviate-the-need-for-testing-and-observability/It&rsquo;s easy to prototype with AI, but production-grade AI apps require even more thorough testing and observability than traditional software.My experience as a Data Tech Lead with Work on Climatehttps://yanirseroussi.com/2024/04/08/my-experience-as-a-data-tech-lead-with-work-on-climate/Mon, 08 Apr 2024 02:00:00 +0000https://yanirseroussi.com/2024/04/08/my-experience-as-a-data-tech-lead-with-work-on-climate/The story of how I joined Work on Climate as a volunteer and became its data tech lead, with lessons applied to consulting &amp; fractional work.Artificial intelligence, automation, and the art of counting fishhttps://yanirseroussi.com/2024/04/01/artificial-intelligence-automation-and-the-art-of-counting-fish/Mon, 01 Apr 2024 06:00:00 +0000https://yanirseroussi.com/2024/04/01/artificial-intelligence-automation-and-the-art-of-counting-fish/Discussing the use of AI to automate underwater marine surveys as an example of the uneven distribution of technological advancement.Questions to consider when using AI for PDF data extractionhttps://yanirseroussi.com/2024/03/11/questions-to-consider-when-using-ai-for-pdf-data-extraction/Mon, 11 Mar 2024 00:00:00 +0000https://yanirseroussi.com/2024/03/11/questions-to-consider-when-using-ai-for-pdf-data-extraction/Discussing considerations that arise when attempting to automate the extraction of structured data from PDFs and similar documents.Two types of startup data problemshttps://yanirseroussi.com/2024/03/04/two-types-of-startup-data-problems/Mon, 04 Mar 2024 02:00:00 +0000https://yanirseroussi.com/2024/03/04/two-types-of-startup-data-problems/Classifying startups as ML-centric or non-ML is a helpful exercise to uncover the data challenges they&rsquo;re likely to face.Avoiding AI complexity: First, write no codehttps://yanirseroussi.com/2024/02/26/avoiding-ai-complexity-first-write-no-code/Mon, 26 Feb 2024 01:45:00 +0000https://yanirseroussi.com/2024/02/26/avoiding-ai-complexity-first-write-no-code/Two stories of getting AI functionality to production, which demonstrate the risks inherent in custom development versus starting with a no-code approach.Building your startup's minimum viable data stackhttps://yanirseroussi.com/2024/02/19/building-your-startups-minimum-viable-data-stack/Mon, 19 Feb 2024 00:00:00 +0000https://yanirseroussi.com/2024/02/19/building-your-startups-minimum-viable-data-stack/First post in a series on building a minimum viable data stack for startups, introducing key definitions, components, and considerations.Nudging ChatGPT to invent books you have no time to readhttps://yanirseroussi.com/2024/02/12/nudging-chatgpt-to-invent-books-you-have-no-time-to-read/Mon, 12 Feb 2024 05:00:00 +0000https://yanirseroussi.com/2024/02/12/nudging-chatgpt-to-invent-books-you-have-no-time-to-read/Getting ChatGPT Plus to elaborate on possible book content and produce a PDF cheatsheet, with the goal of learning about its capabilities.Substance over titles: Your first data hire may be a data scientisthttps://yanirseroussi.com/2024/02/05/substance-over-titles-your-first-data-hire-may-be-a-data-scientist/Mon, 05 Feb 2024 02:45:00 +0000https://yanirseroussi.com/2024/02/05/substance-over-titles-your-first-data-hire-may-be-a-data-scientist/Advice for hiring a startup&rsquo;s first data person: match skills to business needs, consider contractors, and get help from data people.New decade, new tagline: Data & AI for Impacthttps://yanirseroussi.com/2024/01/19/new-decade-new-tagline-data-and-ai-for-impact/Fri, 19 Jan 2024 00:00:00 +0000https://yanirseroussi.com/2024/01/19/new-decade-new-tagline-data-and-ai-for-impact/Shifting focus to &lsquo;Data &amp; AI for Impact&rsquo;, with more startup-related content, increased posting frequency, and deeper audience engagement.Supporting volunteer monitoring of marine biodiversity with modern web and data toolshttps://yanirseroussi.com/2023/11/29/supporting-volunteer-monitoring-of-marine-biodiversity-with-modern-web-and-data-tools/Wed, 29 Nov 2023 02:00:00 +0000https://yanirseroussi.com/2023/11/29/supporting-volunteer-monitoring-of-marine-biodiversity-with-modern-web-and-data-tools/Summarising the work Uri Seroussi and I did to improve Reef Life Survey&rsquo;s Reef Species of the World app.Lessons from reluctant data engineeringhttps://yanirseroussi.com/2023/10/25/lessons-from-reluctant-data-engineering/Wed, 25 Oct 2023 04:45:00 +0000https://yanirseroussi.com/2023/10/25/lessons-from-reluctant-data-engineering/Video and summary of a talk I gave at DataEngBytes Brisbane on what I learned from doing data engineering as part of every data science role I had.My rediscovery of quiet writing on the open webhttps://yanirseroussi.com/2023/08/28/my-rediscovery-of-quiet-writing-on-the-open-web/Mon, 28 Aug 2023 05:30:00 +0000https://yanirseroussi.com/2023/08/28/my-rediscovery-of-quiet-writing-on-the-open-web/Reflections on publishing on this website: Writing publicly to share thoughts and documentation beats chasing views and likes.Was data science a failure mode of software engineering?https://yanirseroussi.com/2023/06/30/was-data-science-a-failure-mode-of-software-engineering/Fri, 30 Jun 2023 00:06:30 +0000https://yanirseroussi.com/2023/06/30/was-data-science-a-failure-mode-of-software-engineering/Yes, data science projects have suffered from classic software engineering mistakes, but the field is maturing with the rise of new engineering roles.How hackable are automated coding assessments?https://yanirseroussi.com/2023/05/26/how-hackable-are-automated-coding-assessments/Fri, 26 May 2023 00:03:00 +0000https://yanirseroussi.com/2023/05/26/how-hackable-are-automated-coding-assessments/Exploring the hackability of speed-based coding tests, using CodeSignal&rsquo;s Industry Coding Framework as a case study.Remaining relevant as a small language modelhttps://yanirseroussi.com/2023/04/21/remaining-relevant-as-a-small-language-model/Fri, 21 Apr 2023 00:06:30 +0000https://yanirseroussi.com/2023/04/21/remaining-relevant-as-a-small-language-model/Bing Chat recently quipped that humans are small language models. Here are some of my thoughts on how we small language models can remain relevant (for now).ChatGPT is transformative AIhttps://yanirseroussi.com/2022/12/11/chatgpt-is-transformative-ai/Sun, 11 Dec 2022 00:00:00 +0000https://yanirseroussi.com/2022/12/11/chatgpt-is-transformative-ai/My perspective after a week of using ChatGPT: This is a step change in finding distilled information, and it&rsquo;s only the beginning.Causal Machine Learning is off to a good start, despite some issueshttps://yanirseroussi.com/2022/09/12/causal-machine-learning-book-draft-review/Mon, 12 Sep 2022 02:45:00 +0000https://yanirseroussi.com/2022/09/12/causal-machine-learning-book-draft-review/Reviewing the first three chapters of the book Causal Machine Learning by Robert Osazuwa Ness.The mission matters: Moving to climate tech as a data scientisthttps://yanirseroussi.com/2022/06/06/the-mission-matters-moving-to-climate-tech-as-a-data-scientist/Mon, 06 Jun 2022 00:00:00 +0000https://yanirseroussi.com/2022/06/06/the-mission-matters-moving-to-climate-tech-as-a-data-scientist/Discussing my recent career move into climate tech as a way of doing more to help mitigate dangerous climate change.Building useful machine learning tools keeps getting easier: A fish ID case studyhttps://yanirseroussi.com/2022/03/20/building-useful-machine-learning-tools-keeps-getting-easier-a-fish-id-case-study/Sun, 20 Mar 2022 04:30:00 +0000https://yanirseroussi.com/2022/03/20/building-useful-machine-learning-tools-keeps-getting-easier-a-fish-id-case-study/Lessons learned building a fish ID web app with fast.ai and Streamlit, in an attempt to reduce my fear of missing out on the latest deep learning developments.Analysis strategies in online A/B experiments: Intention-to-treat, per-protocol, and other lessons from clinical trialshttps://yanirseroussi.com/2022/01/14/analysis-strategies-in-online-a-b-experiments/Fri, 14 Jan 2022 00:05:40 +0000https://yanirseroussi.com/2022/01/14/analysis-strategies-in-online-a-b-experiments/Epidemiologists analyse clinical trials to estimate the intention-to-treat and per-protocol effects. This post applies their strategies to online experiments.Use your human brain to avoid artificial intelligence disastershttps://yanirseroussi.com/2021/11/22/use-your-human-brain-to-avoid-artificial-intelligence-disasters/Mon, 22 Nov 2021 03:45:00 +0000https://yanirseroussi.com/2021/11/22/use-your-human-brain-to-avoid-artificial-intelligence-disasters/Overview of a talk I gave at a deep learning course, focusing on AI ethics as the need for humans to think on the context and consequences of applying AI.Migrating from WordPress.com to Hugo on GitHub + Cloudflarehttps://yanirseroussi.com/2021/11/10/migrating-from-wordpress-com-to-hugo-on-github-cloudflare/Wed, 10 Nov 2021 06:30:00 +0000https://yanirseroussi.com/2021/11/10/migrating-from-wordpress-com-to-hugo-on-github-cloudflare/My reasons for switching from WordPress.com to Hugo on GitHub + Cloudflare, along with a summary of the solution components and migration process.My work with Automattichttps://yanirseroussi.com/2021/10/07/my-work-with-automattic/Thu, 07 Oct 2021 00:00:00 +0000https://yanirseroussi.com/2021/10/07/my-work-with-automattic/Back-dated meta-post that gathers my posts on Automattic blogs into a summary of the work I&rsquo;ve done with the company.Some highlights from 2020https://yanirseroussi.com/2021/04/05/some-highlights-from-2020/Mon, 05 Apr 2021 06:41:48 +0000https://yanirseroussi.com/2021/04/05/some-highlights-from-2020/Sharing remote teamwork insights, my climate &amp; sustainability activism, Reef Life Survey publications, and progress on Automattic&rsquo;s Experimentation Platform.Many is not enough: Counting simulations to bootstrap the right wayhttps://yanirseroussi.com/2020/08/24/many-is-not-enough-counting-simulations-to-bootstrap-the-right-way/Mon, 24 Aug 2020 01:35:17 +0000https://yanirseroussi.com/2020/08/24/many-is-not-enough-counting-simulations-to-bootstrap-the-right-way/Going deeper into correct testing of different methods for bootstrap estimation of confidence intervals.Software commodities are eating interesting data science workhttps://yanirseroussi.com/2020/01/11/software-commodities-are-eating-interesting-data-science-work/Sat, 11 Jan 2020 09:22:35 +0000https://yanirseroussi.com/2020/01/11/software-commodities-are-eating-interesting-data-science-work/Being a data scientist can sometimes feel like a race against software commodities that replace interesting work. What can one do to remain relevant?A day in the life of a remote data scientisthttps://yanirseroussi.com/2019/12/12/a-day-in-the-life-of-a-remote-data-scientist/Wed, 11 Dec 2019 22:06:19 +0000https://yanirseroussi.com/2019/12/12/a-day-in-the-life-of-a-remote-data-scientist/Video of a talk I gave on remote data science work at the Data Science Sydney meetup.Bootstrapping the right way?https://yanirseroussi.com/2019/10/06/bootstrapping-the-right-way/Sun, 06 Oct 2019 06:48:07 +0000https://yanirseroussi.com/2019/10/06/bootstrapping-the-right-way/Video and summary of a talk I gave at YOW! Data on bootstrap estimation of confidence intervals.Hackers beware: Bootstrap sampling may be harmfulhttps://yanirseroussi.com/2019/01/08/hackers-beware-bootstrap-sampling-may-be-harmful/Mon, 07 Jan 2019 21:07:56 +0000https://yanirseroussi.com/2019/01/08/hackers-beware-bootstrap-sampling-may-be-harmful/Bootstrap sampling has been promoted as an easy way of modelling uncertainty to hackers without much statistical knowledge. But things aren&rsquo;t that simple.The most practical causal inference book I’ve read (is still a draft)https://yanirseroussi.com/2018/12/24/the-most-practical-causal-inference-book-ive-read-is-still-a-draft/Mon, 24 Dec 2018 02:37:50 +0000https://yanirseroussi.com/2018/12/24/the-most-practical-causal-inference-book-ive-read-is-still-a-draft/Causal Inference by Miguel Hernán and Jamie Robins is a must-read for anyone interested in the area.Reflections on remote data science workhttps://yanirseroussi.com/2018/11/03/reflections-on-remote-data-science-work/Sat, 03 Nov 2018 06:33:13 +0000https://yanirseroussi.com/2018/11/03/reflections-on-remote-data-science-work/Discussing the pluses and minuses of remote work eighteen months after joining Automattic as a data scientist.Defining data science in 2018https://yanirseroussi.com/2018/07/22/defining-data-science-in-2018/Sun, 22 Jul 2018 08:27:43 +0000https://yanirseroussi.com/2018/07/22/defining-data-science-in-2018/Updating my definition of data science to match changes in the field. It is now broader than before, but its ultimate goal is still to support decisions.Advice for aspiring data scientists and other FAQshttps://yanirseroussi.com/2017/10/15/advice-for-aspiring-data-scientists-and-other-faqs/Sun, 15 Oct 2017 09:15:25 +0000https://yanirseroussi.com/2017/10/15/advice-for-aspiring-data-scientists-and-other-faqs/Frequently asked questions by visitors to this site, especially around entering the data science field.State of Bandcamp Recommender, Late 2017https://yanirseroussi.com/2017/09/02/state-of-bandcamp-recommender/Sat, 02 Sep 2017 10:19:02 +0000https://yanirseroussi.com/2017/09/02/state-of-bandcamp-recommender/Call for BCRecommender maintainers followed by a decision to shut it down, as I don&rsquo;t have enough time and Bandcamp now offers recommendations.My 10-step path to becoming a remote data scientist with Automattichttps://yanirseroussi.com/2017/07/29/my-10-step-path-to-becoming-a-remote-data-scientist-with-automattic/Sat, 29 Jul 2017 05:39:26 +0000https://yanirseroussi.com/2017/07/29/my-10-step-path-to-becoming-a-remote-data-scientist-with-automattic/I wanted a well-paid data science-y remote job with an established company that offers a good life balance and makes products I care about. I got it eventually.Exploring and visualising Reef Life Survey datahttps://yanirseroussi.com/2017/06/03/exploring-and-visualising-reef-life-survey-data/Sat, 03 Jun 2017 00:49:05 +0000https://yanirseroussi.com/2017/06/03/exploring-and-visualising-reef-life-survey-data/Web tools I built to visualise Reef Life Survey data and assist citizen scientists in underwater visual census work.Customer lifetime value and the proliferation of misinformation on the internethttps://yanirseroussi.com/2017/01/08/customer-lifetime-value-and-the-proliferation-of-misinformation-on-the-internet/Sun, 08 Jan 2017 20:02:30 +0000https://yanirseroussi.com/2017/01/08/customer-lifetime-value-and-the-proliferation-of-misinformation-on-the-internet/There&rsquo;s a lot of misleading content on the estimation of customer lifetime value. Here&rsquo;s what I learned about doing it well.Ask Why! Finding motives, causes, and purpose in data sciencehttps://yanirseroussi.com/2016/09/19/ask-why-finding-motives-causes-and-purpose-in-data-science/Mon, 19 Sep 2016 21:28:44 +0000https://yanirseroussi.com/2016/09/19/ask-why-finding-motives-causes-and-purpose-in-data-science/Video and summary of a talk I gave at the Data Science Sydney meetup, about going beyond the what &amp; how of predictive modelling.If you don’t pay attention, data can drive you off a cliffhttps://yanirseroussi.com/2016/08/21/seven-ways-to-be-data-driven-off-a-cliff/Sun, 21 Aug 2016 21:34:17 +0000https://yanirseroussi.com/2016/08/21/seven-ways-to-be-data-driven-off-a-cliff/Seven common mistakes to avoid when working with data, such as ignoring uncertainty and confusing observed and unobserved quantities.Is Data Scientist a useless job title?https://yanirseroussi.com/2016/08/04/is-data-scientist-a-useless-job-title/Thu, 04 Aug 2016 22:26:03 +0000https://yanirseroussi.com/2016/08/04/is-data-scientist-a-useless-job-title/It seems like anyone who touches data can call themselves a data scientist, which makes the title useless. The work they do can still be useful, though.Making Bayesian A/B testing more accessiblehttps://yanirseroussi.com/2016/06/19/making-bayesian-ab-testing-more-accessible/Sun, 19 Jun 2016 10:32:15 +0000https://yanirseroussi.com/2016/06/19/making-bayesian-ab-testing-more-accessible/A web tool I built to interpret A/B test results in a Bayesian way, including prior specification, visualisations, and decision rules.Diving deeper into causality: Pearl, Kleinberg, Hill, and untested assumptionshttps://yanirseroussi.com/2016/05/15/diving-deeper-into-causality-pearl-kleinberg-hill-and-untested-assumptions/Sat, 14 May 2016 19:57:03 +0000https://yanirseroussi.com/2016/05/15/diving-deeper-into-causality-pearl-kleinberg-hill-and-untested-assumptions/Discussing the need for untested assumptions and temporality in causal inference. Mostly based on Samantha Kleinberg&rsquo;s Causality, Probability, and Time.The rise of greedy robotshttps://yanirseroussi.com/2016/03/20/the-rise-of-greedy-robots/Sun, 20 Mar 2016 20:33:43 +0000https://yanirseroussi.com/2016/03/20/the-rise-of-greedy-robots/Is artificial/machine intelligence a future threat? I argue that it&rsquo;s already here, with greedy robots already dominating our lives.Why you should stop worrying about deep learning and deepen your understanding of causality insteadhttps://yanirseroussi.com/2016/02/14/why-you-should-stop-worrying-about-deep-learning-and-deepen-your-understanding-of-causality-instead/Sun, 14 Feb 2016 11:04:11 +0000https://yanirseroussi.com/2016/02/14/why-you-should-stop-worrying-about-deep-learning-and-deepen-your-understanding-of-causality-instead/Causality is often overlooked but is of much higher relevance to most data scientists than deep learning.The joys of offline data collectionhttps://yanirseroussi.com/2016/01/24/the-joys-of-offline-data-collection/Sun, 24 Jan 2016 00:32:25 +0000https://yanirseroussi.com/2016/01/24/the-joys-of-offline-data-collection/Insights on data collection and machine learning from spending a month sailing, diving, and counting fish with Reef Life Survey.This holiday season, give me real insightshttps://yanirseroussi.com/2015/12/08/this-holiday-season-give-me-real-insights/Tue, 08 Dec 2015 06:57:25 +0000https://yanirseroussi.com/2015/12/08/this-holiday-season-give-me-real-insights/Some companies present raw data or information as &ldquo;insights&rdquo;. This post surveys some examples, and discusses how they can be turned into real insights.The hardest parts of data sciencehttps://yanirseroussi.com/2015/11/23/the-hardest-parts-of-data-science/Mon, 23 Nov 2015 04:14:21 +0000https://yanirseroussi.com/2015/11/23/the-hardest-parts-of-data-science/Defining feasible problems and coming up with reasonable ways of measuring solutions is harder than building accurate models or obtaining clean data.Migrating a simple web application from MongoDB to Elasticsearchhttps://yanirseroussi.com/2015/11/04/migrating-a-simple-web-application-from-mongodb-to-elasticsearch/Wed, 04 Nov 2015 03:53:18 +0000https://yanirseroussi.com/2015/11/04/migrating-a-simple-web-application-from-mongodb-to-elasticsearch/Migrating BCRecommender from MongoDB to Elasticsearch made it possible to offer a richer search experience to users at a similar cost, among other benefits.Miscommunicating science: Simplistic models, nutritionism, and the art of storytellinghttps://yanirseroussi.com/2015/10/19/nutritionism-and-the-need-for-complex-models-to-explain-complex-phenomena/Mon, 19 Oct 2015 00:02:32 +0000https://yanirseroussi.com/2015/10/19/nutritionism-and-the-need-for-complex-models-to-explain-complex-phenomena/Nutritionism is a special case of misinterpretation and miscommunication of scientific results – something many data scientists encounter in their work.The wonderful world of recommender systemshttps://yanirseroussi.com/2015/10/02/the-wonderful-world-of-recommender-systems/Fri, 02 Oct 2015 05:25:57 +0000https://yanirseroussi.com/2015/10/02/the-wonderful-world-of-recommender-systems/Giving an overview of the field and common paradigms, and debunking five common myths about recommender systems.You don’t need a data scientist (yet)https://yanirseroussi.com/2015/08/24/you-dont-need-a-data-scientist-yet/Mon, 24 Aug 2015 08:25:30 +0000https://yanirseroussi.com/2015/08/24/you-dont-need-a-data-scientist-yet/Hiring data scientists prematurely is wasteful and frustrating. Here are some questions to ask before you hire your first data scientist.Goodbye, Parse.comhttps://yanirseroussi.com/2015/07/31/goodbye-parse-com/Fri, 31 Jul 2015 03:29:50 +0000https://yanirseroussi.com/2015/07/31/goodbye-parse-com/Migrating my web apps away from Parse.com due to reliability issues. Self-hosting is a better solution.Learning about deep learning through album cover classificationhttps://yanirseroussi.com/2015/07/06/learning-about-deep-learning-through-album-cover-classification/Mon, 06 Jul 2015 22:21:42 +0000https://yanirseroussi.com/2015/07/06/learning-about-deep-learning-through-album-cover-classification/Progress on my album cover classification project, highlighting lessons that would be useful to others who are getting started with deep learning.Hopping on the deep learning bandwagonhttps://yanirseroussi.com/2015/06/06/hopping-on-the-deep-learning-bandwagon/Sat, 06 Jun 2015 05:00:22 +0000https://yanirseroussi.com/2015/06/06/hopping-on-the-deep-learning-bandwagon/To become proficient at solving data science problems, you need to get your hands dirty. Here, I used album cover classification to learn about deep learning.First steps in data science: author-aware sentiment analysishttps://yanirseroussi.com/2015/05/02/first-steps-in-data-science-author-aware-sentiment-analysis/Sat, 02 May 2015 08:31:10 +0000https://yanirseroussi.com/2015/05/02/first-steps-in-data-science-author-aware-sentiment-analysis/I became a data scientist by doing a PhD, but the same steps can be followed without a formal education program.My divestment from fossil fuelshttps://yanirseroussi.com/2015/04/24/my-divestment-from-fossil-fuels/Fri, 24 Apr 2015 00:19:36 +0000https://yanirseroussi.com/2015/04/24/my-divestment-from-fossil-fuels/Recent choices I&rsquo;ve made to reduce my exposure to fossil fuels, including practical steps that can be taken by Australians and generally applicable lessons.My PhD workhttps://yanirseroussi.com/phd-work/Mon, 30 Mar 2015 03:23:33 +0000https://yanirseroussi.com/phd-work/An overview of my PhD in data science / artificial intelligence. Thesis title: Text Mining and Rating Prediction with Topical User Models.The long road to a lifestyle businesshttps://yanirseroussi.com/2015/03/22/the-long-road-to-a-lifestyle-business/Sun, 22 Mar 2015 09:43:47 +0000https://yanirseroussi.com/2015/03/22/the-long-road-to-a-lifestyle-business/Progress since leaving my last full-time job and setting on an independent path that includes data science consulting and work on my own projects.Learning to rank for personalised search (Yandex Search Personalisation – Kaggle Competition Summary – Part 2)https://yanirseroussi.com/2015/02/11/learning-to-rank-for-personalised-search-yandex-search-personalisation-kaggle-competition-summary-part-2/Wed, 11 Feb 2015 06:34:17 +0000https://yanirseroussi.com/2015/02/11/learning-to-rank-for-personalised-search-yandex-search-personalisation-kaggle-competition-summary-part-2/My team&rsquo;s solution to the Yandex Search Personalisation competition (finished 9th out of 194 teams).Is thinking like a search engine possible? (Yandex search personalisation – Kaggle competition summary – part 1)https://yanirseroussi.com/2015/01/29/is-thinking-like-a-search-engine-possible-yandex-search-personalisation-kaggle-competition-summary-part-1/Thu, 29 Jan 2015 10:37:39 +0000https://yanirseroussi.com/2015/01/29/is-thinking-like-a-search-engine-possible-yandex-search-personalisation-kaggle-competition-summary-part-1/Insights on search personalisation and SEO from participating in a Kaggle competition (finished 9th out of 194 teams).Automating Parse.com bulk data importshttps://yanirseroussi.com/2015/01/15/automating-parse-com-bulk-data-imports/Thu, 15 Jan 2015 04:41:16 +0000https://yanirseroussi.com/2015/01/15/automating-parse-com-bulk-data-imports/A script for importing data into the Parse backend-as-a-service.Stochastic Gradient Boosting: Choosing the Best Number of Iterationshttps://yanirseroussi.com/2014/12/29/stochastic-gradient-boosting-choosing-the-best-number-of-iterations/Mon, 29 Dec 2014 02:30:06 +0000https://yanirseroussi.com/2014/12/29/stochastic-gradient-boosting-choosing-the-best-number-of-iterations/Exploring an approach to choosing the optimal number of iterations in stochastic gradient boosting, following a bug I found in scikit-learn.SEO: Mostly about showing up?https://yanirseroussi.com/2014/12/15/seo-mostly-about-showing-up/Mon, 15 Dec 2014 04:25:25 +0000https://yanirseroussi.com/2014/12/15/seo-mostly-about-showing-up/Increasing SEO traffic to BCRecommender by adding content and opening up more pages for crawling. It turns out that thin content is better than no content.Fitting noise: Forecasting the sale price of bulldozers (Kaggle competition summary)https://yanirseroussi.com/2014/11/19/fitting-noise-forecasting-the-sale-price-of-bulldozers-kaggle-competition-summary/Wed, 19 Nov 2014 09:17:34 +0000https://yanirseroussi.com/2014/11/19/fitting-noise-forecasting-the-sale-price-of-bulldozers-kaggle-competition-summary/Summary of a Kaggle competition to forecast bulldozer sale price, where I finished 9th out of 476 teams.BCRecommender Traction Updatehttps://yanirseroussi.com/2014/11/05/bcrecommender-traction-update/Wed, 05 Nov 2014 02:29:35 +0000https://yanirseroussi.com/2014/11/05/bcrecommender-traction-update/Update on BCRecommender traction using three channels: blogger outreach, search engine optimisation, and content marketing.What is data science?https://yanirseroussi.com/2014/10/23/what-is-data-science/Thu, 23 Oct 2014 03:22:08 +0000https://yanirseroussi.com/2014/10/23/what-is-data-science/Data science has been a hot term in the past few years. Still, there isn&rsquo;t a single definition of the field. This post discusses my favourite definition.Greek Media Monitoring Kaggle competition: My approachhttps://yanirseroussi.com/2014/10/07/greek-media-monitoring-kaggle-competition-my-approach/Tue, 07 Oct 2014 03:21:35 +0000https://yanirseroussi.com/2014/10/07/greek-media-monitoring-kaggle-competition-my-approach/Summary of my approach to the Greek Media Monitoring Kaggle competition, where I finished 6th out of 120 teams.Applying the Traction Book’s Bullseye framework to BCRecommenderhttps://yanirseroussi.com/2014/09/24/applying-the-traction-books-bullseye-framework-to-bcrecommender/Wed, 24 Sep 2014 04:57:39 +0000https://yanirseroussi.com/2014/09/24/applying-the-traction-books-bullseye-framework-to-bcrecommender/Ranking 19 channels with the goal of getting traction for BCRecommender.Bandcamp recommendation and discovery algorithmshttps://yanirseroussi.com/2014/09/19/bandcamp-recommendation-and-discovery-algorithms/Fri, 19 Sep 2014 14:26:55 +0000https://yanirseroussi.com/2014/09/19/bandcamp-recommendation-and-discovery-algorithms/The recommendation backend for my BCRecommender service for personalised Bandcamp music discovery.Building a recommender system on a shoestring budget (or: BCRecommender part 2 – general system layout)https://yanirseroussi.com/2014/09/07/building-a-recommender-system-on-a-shoestring-budget/Sun, 07 Sep 2014 10:48:44 +0000https://yanirseroussi.com/2014/09/07/building-a-recommender-system-on-a-shoestring-budget/Iterating on my BCRecommender service with the goal of keeping costs low while providing a valuable music recommendation service.Building a Bandcamp recommender system (part 1 – motivation)https://yanirseroussi.com/2014/08/30/building-a-bandcamp-recommender-system-part-1-motivation/Sat, 30 Aug 2014 08:11:38 +0000https://yanirseroussi.com/2014/08/30/building-a-bandcamp-recommender-system-part-1-motivation/My motivation behind building BCRecommender, a free recommendation &amp; discovery service for Bandcamp music.How to (almost) win Kaggle competitionshttps://yanirseroussi.com/2014/08/24/how-to-almost-win-kaggle-competitions/Sun, 24 Aug 2014 12:40:53 +0000https://yanirseroussi.com/2014/08/24/how-to-almost-win-kaggle-competitions/Summary of a talk I gave at the Data Science Sydney meetup with ten tips on almost-winning Kaggle competitions.Data’s hierarchy of needshttps://yanirseroussi.com/2014/08/17/datas-hierarchy-of-needs/Sun, 17 Aug 2014 13:09:30 +0000https://yanirseroussi.com/2014/08/17/datas-hierarchy-of-needs/Discussing the hierarchy of needs proposed by Jay Kreps. Key takeaway: Data-driven algorithms &amp; insights can only be as good as the underlying data.Kaggle competition tips and summarieshttps://yanirseroussi.com/kaggle/Sat, 05 Apr 2014 23:46:10 +0000https://yanirseroussi.com/kaggle/Pointers to all my Kaggle advice posts and competition summaries.Kaggle beginner tipshttps://yanirseroussi.com/2014/01/19/kaggle-beginner-tips/Sun, 19 Jan 2014 10:34:28 +0000https://yanirseroussi.com/2014/01/19/kaggle-beginner-tips/First post! An email I sent to members of the Data Science Sydney Meetup with tips on how to get started with Kaggle competitions. \ No newline at end of file diff --git a/sitemap.xml b/sitemap.xml index 40a3ea85d..56ede548a 100644 --- a/sitemap.xml +++ b/sitemap.xml @@ -1 +1 @@ -https://yanirseroussi.com/tags/business/2024-06-03T12:58:00+10:00https://yanirseroussi.com/tags/career/2024-06-03T12:58:00+10:00https://yanirseroussi.com/tags/data-strategy/2024-06-03T12:58:00+10:00https://yanirseroussi.com/2024/06/03/how-to-avoid-startups-with-poor-development-processes/2024-06-03T12:58:00+10:00https://yanirseroussi.com/tags/software-engineering/2024-06-03T12:58:00+10:00https://yanirseroussi.com/tags/startups/2024-06-03T12:58:00+10:00https://yanirseroussi.com/tags/2024-06-03T12:58:00+10:00https://yanirseroussi.com/2024-06-03T12:58:00+10:00https://yanirseroussi.com/tags/artificial-intelligence/2024-05-27T12:25:30+10:00https://yanirseroussi.com/tags/data-engineering/2024-05-27T12:25:30+10:00https://yanirseroussi.com/tags/data-science/2024-05-27T12:25:30+10:00https://yanirseroussi.com/2024/05/27/plumbing-decisions-and-automation-de-hyping-data-and-ai/2024-05-27T12:25:30+10:00https://yanirseroussi.com/til/2024/05/25/adapting-to-the-economy-of-algorithms/2024-05-25T10:00:56+10:00https://yanirseroussi.com/tags/books/2024-05-25T10:00:56+10:00https://yanirseroussi.com/tags/futurism/2024-05-25T10:00:56+10:00https://yanirseroussi.com/tags/quotes/2024-05-25T10:00:56+10:00https://yanirseroussi.com/2024/05/20/question-startup-culture-before-accepting-a-data-to-ai-role/2024-05-21T17:08:32+10:00https://yanirseroussi.com/2024/05/13/probing-the-people-aspects-of-an-early-stage-startup/2024-05-13T12:41:01+10:00https://yanirseroussi.com/2024/05/06/business-questions-to-ask-before-taking-a-startup-data-role/2024-05-06T14:41:43+10:00https://yanirseroussi.com/tags/consulting/2024-04-29T17:25:28+10:00https://yanirseroussi.com/2024/04/29/mentorship-and-the-art-of-actionable-advice/2024-04-29T17:25:28+10:00https://yanirseroussi.com/tags/personal/2024-04-29T17:25:28+10:00https://yanirseroussi.com/tags/analytics/2024-05-06T16:35:22+10:00https://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/2024-04-22T17:38:21+10:00https://yanirseroussi.com/tags/machine-learning/2024-04-22T17:38:21+10:00https://yanirseroussi.com/2024/04/15/ai-does-not-obviate-the-need-for-testing-and-observability/2024-04-15T15:54:17+10:00https://yanirseroussi.com/tags/linkedin/2024-04-11T13:42:58+10:00https://yanirseroussi.com/til/2024/04/11/linkedin-is-a-teachable-skill/2024-04-11T13:42:58+10:00https://yanirseroussi.com/tags/marketing/2024-04-11T13:42:58+10:00https://yanirseroussi.com/tags/climate-change/2024-04-08T12:13:47+10:00https://yanirseroussi.com/tags/environment/2024-04-08T12:13:47+10:00https://yanirseroussi.com/2024/04/08/my-experience-as-a-data-tech-lead-with-work-on-climate/2024-04-08T12:13:47+10:00https://yanirseroussi.com/tags/remote-work/2024-04-08T12:13:47+10:00https://yanirseroussi.com/til/2024/04/05/the-data-engineering-lifecycle-is-not-going-anywhere/2024-04-05T11:23:38+10:00https://yanirseroussi.com/2024/04/01/artificial-intelligence-automation-and-the-art-of-counting-fish/2024-04-01T17:02:44+10:00https://yanirseroussi.com/tags/marine-science/2024-04-01T17:02:44+10:00https://yanirseroussi.com/tags/reef-life-survey/2024-04-01T17:02:44+10:00https://yanirseroussi.com/til/2024/03/12/atomic-habits-is-full-of-actionable-advice/2024-03-12T16:33:48+10:00https://yanirseroussi.com/tags/productivity/2024-03-12T16:33:48+10:00https://yanirseroussi.com/2024/03/11/questions-to-consider-when-using-ai-for-pdf-data-extraction/2024-03-11T15:53:13+10:00https://yanirseroussi.com/2024/03/04/two-types-of-startup-data-problems/2024-03-05T08:47:19+10:00https://yanirseroussi.com/2024/02/26/avoiding-ai-complexity-first-write-no-code/2024-03-04T12:39:10+10:00https://yanirseroussi.com/2024/02/19/building-your-startups-minimum-viable-data-stack/2024-02-19T11:25:54+10:00https://yanirseroussi.com/til/2024/02/17/the-three-cs-of-indie-consulting-confidence-cash-and-connections/2024-02-17T12:34:00+10:00https://yanirseroussi.com/2024/02/12/nudging-chatgpt-to-invent-books-you-have-no-time-to-read/2024-02-13T08:24:54+10:00https://yanirseroussi.com/til/2024/02/06/future-software-development-may-require-fewer-humans/2024-02-06T16:39:35+10:00https://yanirseroussi.com/2024/02/05/substance-over-titles-your-first-data-hire-may-be-a-data-scientist/2024-02-19T11:25:54+10:00https://yanirseroussi.com/tags/blogging/2024-01-19T16:35:09+10:00https://yanirseroussi.com/2024/01/19/new-decade-new-tagline-data-and-ai-for-impact/2024-01-19T16:35:09+10:00https://yanirseroussi.com/til/2024/01/09/psychographic-specialisations-may-work-for-discipline-generalists/2024-01-09T13:23:28+10:00https://yanirseroussi.com/til/2024/01/08/the-power-of-parasocial-relationships/2024-01-08T16:31:22+10:00https://yanirseroussi.com/tags/data-business/2024-01-16T09:56:03+10:00https://yanirseroussi.com/til/2023/12/18/positioning-is-a-common-problem-for-data-scientists/2023-12-18T10:38:56+10:00https://yanirseroussi.com/tags/energy-markets/2023-12-14T10:46:41+10:00https://yanirseroussi.com/til/2023/12/14/transfer-learning-applies-to-energy-market-bidding/2023-12-14T10:46:41+10:00https://yanirseroussi.com/tags/data-visualisation/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2023/11/29/supporting-volunteer-monitoring-of-marine-biodiversity-with-modern-web-and-data-tools/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/web-development/2024-01-16T09:56:03+10:00https://yanirseroussi.com/til/2023/11/28/our-blue-machine-is-changing-but-we-are-not-helpless/2024-03-12T16:33:31+10:00https://yanirseroussi.com/til/2023/11/21/you-dont-need-a-proprietary-api-for-static-maps/2023-11-21T16:12:27+10:00https://yanirseroussi.com/2023/10/25/lessons-from-reluctant-data-engineering/2024-01-16T09:56:03+10:00https://yanirseroussi.com/til/2023/10/06/artificial-intelligence-was-a-marketing-term-all-along-just-call-it-automation/2023-10-06T15:11:27+10:00https://yanirseroussi.com/tags/ethics/2024-01-16T09:56:03+10:00https://yanirseroussi.com/til/2023/09/25/the-lines-between-solo-consulting-and-product-building-are-blurry/2023-09-25T11:15:26+10:00https://yanirseroussi.com/til/2023/09/21/googles-rules-of-machine-learning-still-apply-in-the-age-of-large-language-models/2023-09-22T07:54:13+10:00https://yanirseroussi.com/2023/08/28/my-rediscovery-of-quiet-writing-on-the-open-web/2024-01-16T09:56:03+10:00https://yanirseroussi.com/til/2023/08/21/the-minimalist-entrepreneur-is-too-prescriptive-for-me/2024-03-12T16:33:31+10:00https://yanirseroussi.com/til/2023/08/17/revisiting-start-small-stay-small-in-2023-chapter-2/2024-03-12T16:33:31+10:00https://yanirseroussi.com/til/2023/08/16/revisiting-start-small-stay-small-in-2023-chapter-1/2024-03-12T16:33:31+10:00https://yanirseroussi.com/til/2023/08/14/email-notifications-on-public-github-commits/2023-08-14T15:44:21+10:00https://yanirseroussi.com/til/2023/08/11/the-rule-of-thirds-can-probably-be-ignored/2023-08-11T14:35:20+10:00https://yanirseroussi.com/tags/devops/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/github/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/security/2023-07-25T09:30:43+10:00https://yanirseroussi.com/til/2023/07/23/using-yubikey-for-ssh-access/2023-07-25T09:30:43+10:00https://yanirseroussi.com/tags/hugo/2024-01-16T09:56:03+10:00https://yanirseroussi.com/til/2023/07/17/making-a-til-section-with-hugo-and-papermod/2023-07-17T17:18:06+10:00https://yanirseroussi.com/til/2023/07/11/you-cant-save-time/2024-03-12T16:33:31+10:00https://yanirseroussi.com/2023/06/30/was-data-science-a-failure-mode-of-software-engineering/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/hackers/2024-02-13T08:24:54+10:00https://yanirseroussi.com/2023/05/26/how-hackable-are-automated-coding-assessments/2024-02-13T08:24:54+10:00https://yanirseroussi.com/tags/machine-intelligence/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2023/04/21/remaining-relevant-as-a-small-language-model/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2022/12/11/chatgpt-is-transformative-ai/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/causal-inference/2024-02-21T11:52:55+10:00https://yanirseroussi.com/2022/09/12/causal-machine-learning-book-draft-review/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/automattic/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/orkestra/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/politics/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/sustainability/2024-02-21T11:52:55+10:00https://yanirseroussi.com/2022/06/06/the-mission-matters-moving-to-climate-tech-as-a-data-scientist/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2022/03/20/building-useful-machine-learning-tools-keeps-getting-easier-a-fish-id-case-study/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/deep-learning/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/fast.ai/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2022/01/14/analysis-strategies-in-online-a-b-experiments/2024-02-21T11:52:55+10:00https://yanirseroussi.com/tags/split-testing/2024-02-21T11:52:55+10:00https://yanirseroussi.com/tags/statistics/2024-05-06T16:35:22+10:00https://yanirseroussi.com/2021/11/22/use-your-human-brain-to-avoid-artificial-intelligence-disasters/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/cloudflare/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2021/11/10/migrating-from-wordpress-com-to-hugo-on-github-cloudflare/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/wordpress/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2021/10/07/my-work-with-automattic/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2021/04/05/some-highlights-from-2020/2024-02-21T11:52:55+10:00https://yanirseroussi.com/tags/bootstrapping/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/confidence-intervals/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2020/08/24/many-is-not-enough-counting-simulations-to-bootstrap-the-right-way/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2020/01/11/software-commodities-are-eating-interesting-data-science-work/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2019/12/12/a-day-in-the-life-of-a-remote-data-scientist/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2019/10/06/bootstrapping-the-right-way/2024-05-06T16:35:22+10:00https://yanirseroussi.com/2019/01/08/hackers-beware-bootstrap-sampling-may-be-harmful/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2018/12/24/the-most-practical-causal-inference-book-ive-read-is-still-a-draft/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2018/11/03/reflections-on-remote-data-science-work/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2018/07/22/defining-data-science-in-2018/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2017/10/15/advice-for-aspiring-data-scientists-and-other-faqs/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/frequently-asked-questions/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/bandcamp/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/bcrecommender/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2017/09/02/state-of-bandcamp-recommender/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/elasticsearch/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2017/07/29/my-10-step-path-to-becoming-a-remote-data-scientist-with-automattic/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2017/06/03/exploring-and-visualising-reef-life-survey-data/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/javascript/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2017/01/08/customer-lifetime-value-and-the-proliferation-of-misinformation-on-the-internet/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/predictive-modelling/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/science-communication/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/search-engine-optimisation/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2016/09/19/ask-why-finding-motives-causes-and-purpose-in-data-science/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/insights/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2016/08/21/seven-ways-to-be-data-driven-off-a-cliff/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2016/08/04/is-data-scientist-a-useless-job-title/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2016/06/19/making-bayesian-ab-testing-more-accessible/2024-02-21T11:52:55+10:00https://yanirseroussi.com/2016/05/15/diving-deeper-into-causality-pearl-kleinberg-hill-and-untested-assumptions/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/economics/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2016/03/20/the-rise-of-greedy-robots/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2016/02/14/why-you-should-stop-worrying-about-deep-learning-and-deepen-your-understanding-of-causality-instead/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/scuba-diving/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2016/01/24/the-joys-of-offline-data-collection/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/facebook/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2015/12/08/this-holiday-season-give-me-real-insights/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/kaggle/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2015/11/23/the-hardest-parts-of-data-science/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2015/11/04/migrating-a-simple-web-application-from-mongodb-to-elasticsearch/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/mongodb/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/health/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2015/10/19/nutritionism-and-the-need-for-complex-models-to-explain-complex-phenomena/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/nutrition/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/nutritionism/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/recommender-systems/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2015/10/02/the-wonderful-world-of-recommender-systems/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2015/08/24/you-dont-need-a-data-scientist-yet/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2015/07/31/goodbye-parse-com/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/parse.com/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2015/07/06/learning-about-deep-learning-through-album-cover-classification/2024-01-16T09:56:03+10:00https://yanirseroussi.com/deep-learning-resources/2021-11-09T15:38:25+10:00https://yanirseroussi.com/2015/06/06/hopping-on-the-deep-learning-bandwagon/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2015/05/02/first-steps-in-data-science-author-aware-sentiment-analysis/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/sentiment-analysis/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/divestment/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/fossil-fuels/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2015/04/24/my-divestment-from-fossil-fuels/2024-01-16T09:56:03+10:00https://yanirseroussi.com/phd-work/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2015/03/22/the-long-road-to-a-lifestyle-business/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/gradient-boosting/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/kaggle-competition/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2015/02/11/learning-to-rank-for-personalised-search-yandex-search-personalisation-kaggle-competition-summary-part-2/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2015/01/29/is-thinking-like-a-search-engine-possible-yandex-search-personalisation-kaggle-competition-summary-part-1/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2015/01/15/automating-parse-com-bulk-data-imports/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/phantomjs/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/scikit-learn/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2014/12/29/stochastic-gradient-boosting-choosing-the-best-number-of-iterations/2023-07-06T09:28:02+10:00https://yanirseroussi.com/2014/12/15/seo-mostly-about-showing-up/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/traction-book/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2014/11/19/fitting-noise-forecasting-the-sale-price-of-bulldozers-kaggle-competition-summary/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/price-forecasting/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2014/11/05/bcrecommender-traction-update/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/music/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2014/10/23/what-is-data-science/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2014/10/07/greek-media-monitoring-kaggle-competition-my-approach/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/multi-label-classification/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2014/09/24/applying-the-traction-books-bullseye-framework-to-bcrecommender/2023-07-06T09:28:02+10:00https://yanirseroussi.com/2014/09/19/bandcamp-recommendation-and-discovery-algorithms/2023-07-06T09:28:02+10:00https://yanirseroussi.com/2014/09/07/building-a-recommender-system-on-a-shoestring-budget/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2014/08/30/building-a-bandcamp-recommender-system-part-1-motivation/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/music-industry/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2014/08/24/how-to-almost-win-kaggle-competitions/2023-07-06T09:28:02+10:00https://yanirseroussi.com/tags/kaggle-beginners/2023-07-06T09:28:02+10:00https://yanirseroussi.com/2014/08/17/datas-hierarchy-of-needs/2024-01-16T09:56:03+10:00https://yanirseroussi.com/kaggle/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2014/01/19/kaggle-beginner-tips/2023-07-06T09:28:02+10:00https://yanirseroussi.com/about/2024-03-08T11:21:16+10:00https://yanirseroussi.com/free-intro-call/2024-05-22T17:54:36+10:00https://yanirseroussi.com/posts/2024-05-09T10:03:31+10:00https://yanirseroussi.com/causal-inference-resources/2023-07-06T16:01:57+10:00https://yanirseroussi.com/consult/2024-05-23T15:31:11+10:00https://yanirseroussi.com/data-to-ai-health-check/2024-05-22T17:53:56+10:00https://yanirseroussi.com/contact/2024-05-23T15:31:11+10:00https://yanirseroussi.com/talks/2024-05-06T16:35:22+10:00https://yanirseroussi.com/til/2024-05-09T10:03:31+10:00 \ No newline at end of file +https://yanirseroussi.com/tags/analytics/2024-06-10T14:23:12+10:00https://yanirseroussi.com/tags/business/2024-06-10T14:23:12+10:00https://yanirseroussi.com/tags/data-science/2024-06-10T14:23:12+10:00https://yanirseroussi.com/tags/data-strategy/2024-06-10T14:23:12+10:00https://yanirseroussi.com/2024/06/10/startup-data-health-starts-with-healthy-event-tracking/2024-06-10T14:23:12+10:00https://yanirseroussi.com/tags/startups/2024-06-10T14:23:12+10:00https://yanirseroussi.com/tags/2024-06-10T14:23:12+10:00https://yanirseroussi.com/2024-06-10T14:23:12+10:00https://yanirseroussi.com/tags/career/2024-06-03T12:58:00+10:00https://yanirseroussi.com/2024/06/03/how-to-avoid-startups-with-poor-development-processes/2024-06-03T12:58:00+10:00https://yanirseroussi.com/tags/software-engineering/2024-06-03T12:58:00+10:00https://yanirseroussi.com/tags/artificial-intelligence/2024-05-27T12:25:30+10:00https://yanirseroussi.com/tags/data-engineering/2024-05-27T12:25:30+10:00https://yanirseroussi.com/2024/05/27/plumbing-decisions-and-automation-de-hyping-data-and-ai/2024-05-27T12:25:30+10:00https://yanirseroussi.com/til/2024/05/25/adapting-to-the-economy-of-algorithms/2024-05-25T10:00:56+10:00https://yanirseroussi.com/tags/books/2024-05-25T10:00:56+10:00https://yanirseroussi.com/tags/futurism/2024-05-25T10:00:56+10:00https://yanirseroussi.com/tags/quotes/2024-05-25T10:00:56+10:00https://yanirseroussi.com/2024/05/20/question-startup-culture-before-accepting-a-data-to-ai-role/2024-05-21T17:08:32+10:00https://yanirseroussi.com/2024/05/13/probing-the-people-aspects-of-an-early-stage-startup/2024-05-13T12:41:01+10:00https://yanirseroussi.com/2024/05/06/business-questions-to-ask-before-taking-a-startup-data-role/2024-05-06T14:41:43+10:00https://yanirseroussi.com/tags/consulting/2024-04-29T17:25:28+10:00https://yanirseroussi.com/2024/04/29/mentorship-and-the-art-of-actionable-advice/2024-04-29T17:25:28+10:00https://yanirseroussi.com/tags/personal/2024-04-29T17:25:28+10:00https://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/2024-04-22T17:38:21+10:00https://yanirseroussi.com/tags/machine-learning/2024-04-22T17:38:21+10:00https://yanirseroussi.com/2024/04/15/ai-does-not-obviate-the-need-for-testing-and-observability/2024-04-15T15:54:17+10:00https://yanirseroussi.com/tags/linkedin/2024-04-11T13:42:58+10:00https://yanirseroussi.com/til/2024/04/11/linkedin-is-a-teachable-skill/2024-04-11T13:42:58+10:00https://yanirseroussi.com/tags/marketing/2024-04-11T13:42:58+10:00https://yanirseroussi.com/tags/climate-change/2024-04-08T12:13:47+10:00https://yanirseroussi.com/tags/environment/2024-04-08T12:13:47+10:00https://yanirseroussi.com/2024/04/08/my-experience-as-a-data-tech-lead-with-work-on-climate/2024-04-08T12:13:47+10:00https://yanirseroussi.com/tags/remote-work/2024-04-08T12:13:47+10:00https://yanirseroussi.com/til/2024/04/05/the-data-engineering-lifecycle-is-not-going-anywhere/2024-04-05T11:23:38+10:00https://yanirseroussi.com/2024/04/01/artificial-intelligence-automation-and-the-art-of-counting-fish/2024-04-01T17:02:44+10:00https://yanirseroussi.com/tags/marine-science/2024-04-01T17:02:44+10:00https://yanirseroussi.com/tags/reef-life-survey/2024-04-01T17:02:44+10:00https://yanirseroussi.com/til/2024/03/12/atomic-habits-is-full-of-actionable-advice/2024-03-12T16:33:48+10:00https://yanirseroussi.com/tags/productivity/2024-03-12T16:33:48+10:00https://yanirseroussi.com/2024/03/11/questions-to-consider-when-using-ai-for-pdf-data-extraction/2024-03-11T15:53:13+10:00https://yanirseroussi.com/2024/03/04/two-types-of-startup-data-problems/2024-03-05T08:47:19+10:00https://yanirseroussi.com/2024/02/26/avoiding-ai-complexity-first-write-no-code/2024-03-04T12:39:10+10:00https://yanirseroussi.com/2024/02/19/building-your-startups-minimum-viable-data-stack/2024-02-19T11:25:54+10:00https://yanirseroussi.com/til/2024/02/17/the-three-cs-of-indie-consulting-confidence-cash-and-connections/2024-02-17T12:34:00+10:00https://yanirseroussi.com/2024/02/12/nudging-chatgpt-to-invent-books-you-have-no-time-to-read/2024-02-13T08:24:54+10:00https://yanirseroussi.com/til/2024/02/06/future-software-development-may-require-fewer-humans/2024-02-06T16:39:35+10:00https://yanirseroussi.com/2024/02/05/substance-over-titles-your-first-data-hire-may-be-a-data-scientist/2024-02-19T11:25:54+10:00https://yanirseroussi.com/tags/blogging/2024-01-19T16:35:09+10:00https://yanirseroussi.com/2024/01/19/new-decade-new-tagline-data-and-ai-for-impact/2024-01-19T16:35:09+10:00https://yanirseroussi.com/til/2024/01/09/psychographic-specialisations-may-work-for-discipline-generalists/2024-01-09T13:23:28+10:00https://yanirseroussi.com/til/2024/01/08/the-power-of-parasocial-relationships/2024-01-08T16:31:22+10:00https://yanirseroussi.com/tags/data-business/2024-01-16T09:56:03+10:00https://yanirseroussi.com/til/2023/12/18/positioning-is-a-common-problem-for-data-scientists/2023-12-18T10:38:56+10:00https://yanirseroussi.com/tags/energy-markets/2023-12-14T10:46:41+10:00https://yanirseroussi.com/til/2023/12/14/transfer-learning-applies-to-energy-market-bidding/2023-12-14T10:46:41+10:00https://yanirseroussi.com/tags/data-visualisation/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2023/11/29/supporting-volunteer-monitoring-of-marine-biodiversity-with-modern-web-and-data-tools/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/web-development/2024-01-16T09:56:03+10:00https://yanirseroussi.com/til/2023/11/28/our-blue-machine-is-changing-but-we-are-not-helpless/2024-03-12T16:33:31+10:00https://yanirseroussi.com/til/2023/11/21/you-dont-need-a-proprietary-api-for-static-maps/2023-11-21T16:12:27+10:00https://yanirseroussi.com/2023/10/25/lessons-from-reluctant-data-engineering/2024-01-16T09:56:03+10:00https://yanirseroussi.com/til/2023/10/06/artificial-intelligence-was-a-marketing-term-all-along-just-call-it-automation/2023-10-06T15:11:27+10:00https://yanirseroussi.com/tags/ethics/2024-01-16T09:56:03+10:00https://yanirseroussi.com/til/2023/09/25/the-lines-between-solo-consulting-and-product-building-are-blurry/2023-09-25T11:15:26+10:00https://yanirseroussi.com/til/2023/09/21/googles-rules-of-machine-learning-still-apply-in-the-age-of-large-language-models/2023-09-22T07:54:13+10:00https://yanirseroussi.com/2023/08/28/my-rediscovery-of-quiet-writing-on-the-open-web/2024-01-16T09:56:03+10:00https://yanirseroussi.com/til/2023/08/21/the-minimalist-entrepreneur-is-too-prescriptive-for-me/2024-03-12T16:33:31+10:00https://yanirseroussi.com/til/2023/08/17/revisiting-start-small-stay-small-in-2023-chapter-2/2024-03-12T16:33:31+10:00https://yanirseroussi.com/til/2023/08/16/revisiting-start-small-stay-small-in-2023-chapter-1/2024-03-12T16:33:31+10:00https://yanirseroussi.com/til/2023/08/14/email-notifications-on-public-github-commits/2023-08-14T15:44:21+10:00https://yanirseroussi.com/til/2023/08/11/the-rule-of-thirds-can-probably-be-ignored/2023-08-11T14:35:20+10:00https://yanirseroussi.com/tags/devops/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/github/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/security/2023-07-25T09:30:43+10:00https://yanirseroussi.com/til/2023/07/23/using-yubikey-for-ssh-access/2023-07-25T09:30:43+10:00https://yanirseroussi.com/tags/hugo/2024-01-16T09:56:03+10:00https://yanirseroussi.com/til/2023/07/17/making-a-til-section-with-hugo-and-papermod/2023-07-17T17:18:06+10:00https://yanirseroussi.com/til/2023/07/11/you-cant-save-time/2024-03-12T16:33:31+10:00https://yanirseroussi.com/2023/06/30/was-data-science-a-failure-mode-of-software-engineering/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/hackers/2024-02-13T08:24:54+10:00https://yanirseroussi.com/2023/05/26/how-hackable-are-automated-coding-assessments/2024-02-13T08:24:54+10:00https://yanirseroussi.com/tags/machine-intelligence/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2023/04/21/remaining-relevant-as-a-small-language-model/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2022/12/11/chatgpt-is-transformative-ai/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/causal-inference/2024-02-21T11:52:55+10:00https://yanirseroussi.com/2022/09/12/causal-machine-learning-book-draft-review/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/automattic/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/orkestra/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/politics/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/sustainability/2024-02-21T11:52:55+10:00https://yanirseroussi.com/2022/06/06/the-mission-matters-moving-to-climate-tech-as-a-data-scientist/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2022/03/20/building-useful-machine-learning-tools-keeps-getting-easier-a-fish-id-case-study/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/deep-learning/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/fast.ai/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2022/01/14/analysis-strategies-in-online-a-b-experiments/2024-02-21T11:52:55+10:00https://yanirseroussi.com/tags/split-testing/2024-02-21T11:52:55+10:00https://yanirseroussi.com/tags/statistics/2024-05-06T16:35:22+10:00https://yanirseroussi.com/2021/11/22/use-your-human-brain-to-avoid-artificial-intelligence-disasters/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/cloudflare/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2021/11/10/migrating-from-wordpress-com-to-hugo-on-github-cloudflare/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/wordpress/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2021/10/07/my-work-with-automattic/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2021/04/05/some-highlights-from-2020/2024-02-21T11:52:55+10:00https://yanirseroussi.com/tags/bootstrapping/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/confidence-intervals/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2020/08/24/many-is-not-enough-counting-simulations-to-bootstrap-the-right-way/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2020/01/11/software-commodities-are-eating-interesting-data-science-work/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2019/12/12/a-day-in-the-life-of-a-remote-data-scientist/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2019/10/06/bootstrapping-the-right-way/2024-05-06T16:35:22+10:00https://yanirseroussi.com/2019/01/08/hackers-beware-bootstrap-sampling-may-be-harmful/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2018/12/24/the-most-practical-causal-inference-book-ive-read-is-still-a-draft/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2018/11/03/reflections-on-remote-data-science-work/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2018/07/22/defining-data-science-in-2018/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2017/10/15/advice-for-aspiring-data-scientists-and-other-faqs/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/frequently-asked-questions/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/bandcamp/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/bcrecommender/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2017/09/02/state-of-bandcamp-recommender/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/elasticsearch/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2017/07/29/my-10-step-path-to-becoming-a-remote-data-scientist-with-automattic/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2017/06/03/exploring-and-visualising-reef-life-survey-data/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/javascript/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2017/01/08/customer-lifetime-value-and-the-proliferation-of-misinformation-on-the-internet/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/predictive-modelling/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/science-communication/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/search-engine-optimisation/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2016/09/19/ask-why-finding-motives-causes-and-purpose-in-data-science/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/insights/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2016/08/21/seven-ways-to-be-data-driven-off-a-cliff/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2016/08/04/is-data-scientist-a-useless-job-title/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2016/06/19/making-bayesian-ab-testing-more-accessible/2024-02-21T11:52:55+10:00https://yanirseroussi.com/2016/05/15/diving-deeper-into-causality-pearl-kleinberg-hill-and-untested-assumptions/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/economics/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2016/03/20/the-rise-of-greedy-robots/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2016/02/14/why-you-should-stop-worrying-about-deep-learning-and-deepen-your-understanding-of-causality-instead/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/scuba-diving/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2016/01/24/the-joys-of-offline-data-collection/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/facebook/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2015/12/08/this-holiday-season-give-me-real-insights/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/kaggle/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2015/11/23/the-hardest-parts-of-data-science/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2015/11/04/migrating-a-simple-web-application-from-mongodb-to-elasticsearch/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/mongodb/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/health/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2015/10/19/nutritionism-and-the-need-for-complex-models-to-explain-complex-phenomena/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/nutrition/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/nutritionism/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/recommender-systems/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2015/10/02/the-wonderful-world-of-recommender-systems/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2015/08/24/you-dont-need-a-data-scientist-yet/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2015/07/31/goodbye-parse-com/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/parse.com/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2015/07/06/learning-about-deep-learning-through-album-cover-classification/2024-01-16T09:56:03+10:00https://yanirseroussi.com/deep-learning-resources/2021-11-09T15:38:25+10:00https://yanirseroussi.com/2015/06/06/hopping-on-the-deep-learning-bandwagon/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2015/05/02/first-steps-in-data-science-author-aware-sentiment-analysis/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/sentiment-analysis/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/divestment/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/fossil-fuels/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2015/04/24/my-divestment-from-fossil-fuels/2024-01-16T09:56:03+10:00https://yanirseroussi.com/phd-work/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2015/03/22/the-long-road-to-a-lifestyle-business/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/gradient-boosting/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/kaggle-competition/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2015/02/11/learning-to-rank-for-personalised-search-yandex-search-personalisation-kaggle-competition-summary-part-2/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2015/01/29/is-thinking-like-a-search-engine-possible-yandex-search-personalisation-kaggle-competition-summary-part-1/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2015/01/15/automating-parse-com-bulk-data-imports/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/phantomjs/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/scikit-learn/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2014/12/29/stochastic-gradient-boosting-choosing-the-best-number-of-iterations/2023-07-06T09:28:02+10:00https://yanirseroussi.com/2014/12/15/seo-mostly-about-showing-up/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/traction-book/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2014/11/19/fitting-noise-forecasting-the-sale-price-of-bulldozers-kaggle-competition-summary/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/price-forecasting/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2014/11/05/bcrecommender-traction-update/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/music/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2014/10/23/what-is-data-science/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2014/10/07/greek-media-monitoring-kaggle-competition-my-approach/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/multi-label-classification/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2014/09/24/applying-the-traction-books-bullseye-framework-to-bcrecommender/2023-07-06T09:28:02+10:00https://yanirseroussi.com/2014/09/19/bandcamp-recommendation-and-discovery-algorithms/2023-07-06T09:28:02+10:00https://yanirseroussi.com/2014/09/07/building-a-recommender-system-on-a-shoestring-budget/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2014/08/30/building-a-bandcamp-recommender-system-part-1-motivation/2024-01-16T09:56:03+10:00https://yanirseroussi.com/tags/music-industry/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2014/08/24/how-to-almost-win-kaggle-competitions/2023-07-06T09:28:02+10:00https://yanirseroussi.com/tags/kaggle-beginners/2023-07-06T09:28:02+10:00https://yanirseroussi.com/2014/08/17/datas-hierarchy-of-needs/2024-01-16T09:56:03+10:00https://yanirseroussi.com/kaggle/2024-01-16T09:56:03+10:00https://yanirseroussi.com/2014/01/19/kaggle-beginner-tips/2023-07-06T09:28:02+10:00https://yanirseroussi.com/about/2024-03-08T11:21:16+10:00https://yanirseroussi.com/free-intro-call/2024-05-22T17:54:36+10:00https://yanirseroussi.com/posts/2024-05-09T10:03:31+10:00https://yanirseroussi.com/causal-inference-resources/2023-07-06T16:01:57+10:00https://yanirseroussi.com/consult/2024-05-23T15:31:11+10:00https://yanirseroussi.com/data-to-ai-health-check/2024-05-22T17:53:56+10:00https://yanirseroussi.com/contact/2024-05-23T15:31:11+10:00https://yanirseroussi.com/talks/2024-05-06T16:35:22+10:00https://yanirseroussi.com/til/2024-05-09T10:03:31+10:00 \ No newline at end of file diff --git a/tags/analytics/index.html b/tags/analytics/index.html index 7376bd94a..400943f4d 100644 --- a/tags/analytics/index.html +++ b/tags/analytics/index.html @@ -2,7 +2,7 @@

      Analytics

      Assessing a startup's data-to-AI health

      Reviewing the areas that should be assessed to determine a startup’s opportunities and challenges on the data/AI/ML front.

      April 22, 2024

      Substance over titles: Your first data hire may be a data scientist

      Advice for hiring a startup’s first data person: match skills to business needs, consider contractors, and get help from data people.

      February 5, 2024

      Bootstrapping the right way?

      Video and summary of a talk I gave at YOW! Data on bootstrap estimation of confidence intervals.

      October 6, 2019

      Defining data science in 2018

      Updating my definition of data science to match changes in the field. It is now broader than before, but its ultimate goal is still to support decisions.

      July 22, 2018

      Customer lifetime value and the proliferation of misinformation on the internet

      There’s a lot of misleading content on the estimation of customer lifetime value. Here’s what I learned about doing it well.

      January 8, 2017

      If you don’t pay attention, data can drive you off a cliff

      Seven common mistakes to avoid when working with data, such as ignoring uncertainty and confusing observed and unobserved quantities.

      August 21, 2016

      Making Bayesian A/B testing more accessible

      A web tool I built to interpret A/B test results in a Bayesian way, including prior specification, visualisations, and decision rules.

      June 19, 2016

      Why you should stop worrying about deep learning and deepen your understanding of causality instead

      Causality is often overlooked but is of much higher relevance to most data scientists than deep learning.

      February 14, 2016

      This holiday season, give me real insights

      Some companies present raw data or information as “insights”. This post surveys some examples, and discusses how they can be turned into real insights.

      December 8, 2015
      Text and figures licensed under CC BY-NC-ND 4.0 by Yanir Seroussi, except where noted otherwise  | +">

      Analytics

      Startup data health starts with healthy event tracking

      Expanding on the startup health check question of tracking Kukuyeva’s five business aspects as wide events.

      June 10, 2024

      Assessing a startup's data-to-AI health

      Reviewing the areas that should be assessed to determine a startup’s opportunities and challenges on the data/AI/ML front.

      April 22, 2024

      Substance over titles: Your first data hire may be a data scientist

      Advice for hiring a startup’s first data person: match skills to business needs, consider contractors, and get help from data people.

      February 5, 2024

      Bootstrapping the right way?

      Video and summary of a talk I gave at YOW! Data on bootstrap estimation of confidence intervals.

      October 6, 2019

      Defining data science in 2018

      Updating my definition of data science to match changes in the field. It is now broader than before, but its ultimate goal is still to support decisions.

      July 22, 2018

      Customer lifetime value and the proliferation of misinformation on the internet

      There’s a lot of misleading content on the estimation of customer lifetime value. Here’s what I learned about doing it well.

      January 8, 2017

      If you don’t pay attention, data can drive you off a cliff

      Seven common mistakes to avoid when working with data, such as ignoring uncertainty and confusing observed and unobserved quantities.

      August 21, 2016

      Making Bayesian A/B testing more accessible

      A web tool I built to interpret A/B test results in a Bayesian way, including prior specification, visualisations, and decision rules.

      June 19, 2016

      Why you should stop worrying about deep learning and deepen your understanding of causality instead

      Causality is often overlooked but is of much higher relevance to most data scientists than deep learning.

      February 14, 2016

      This holiday season, give me real insights

      Some companies present raw data or information as “insights”. This post surveys some examples, and discusses how they can be turned into real insights.

      December 8, 2015
      Text and figures licensed under CC BY-NC-ND 4.0 by Yanir Seroussi, except where noted otherwise  | Powered by Hugo & PaperMod
      \ No newline at end of file diff --git a/tags/analytics/index.xml b/tags/analytics/index.xml index 25027320b..95bc97f88 100644 --- a/tags/analytics/index.xml +++ b/tags/analytics/index.xml @@ -1 +1 @@ -Analytics on Yanir Seroussi | Data & AI for Startup Impacthttps://yanirseroussi.com/tags/analytics/Recent content in Analytics on Yanir Seroussi | Data & AI for Startup ImpactHugo -- gohugo.ioen-auText and figures licensed under [CC BY-NC-ND 4.0](https://creativecommons.org/licenses/by-nc-nd/4.0/) by [Yanir Seroussi](https://yanirseroussi.com/about/), except where noted otherwiseMon, 22 Apr 2024 06:00:00 +0000Assessing a startup's data-to-AI healthhttps://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/Mon, 22 Apr 2024 06:00:00 +0000https://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/Reviewing the areas that should be assessed to determine a startup&rsquo;s opportunities and challenges on the data/AI/ML front.Substance over titles: Your first data hire may be a data scientisthttps://yanirseroussi.com/2024/02/05/substance-over-titles-your-first-data-hire-may-be-a-data-scientist/Mon, 05 Feb 2024 02:45:00 +0000https://yanirseroussi.com/2024/02/05/substance-over-titles-your-first-data-hire-may-be-a-data-scientist/Advice for hiring a startup&rsquo;s first data person: match skills to business needs, consider contractors, and get help from data people.Bootstrapping the right way?https://yanirseroussi.com/2019/10/06/bootstrapping-the-right-way/Sun, 06 Oct 2019 06:48:07 +0000https://yanirseroussi.com/2019/10/06/bootstrapping-the-right-way/Video and summary of a talk I gave at YOW! Data on bootstrap estimation of confidence intervals.Defining data science in 2018https://yanirseroussi.com/2018/07/22/defining-data-science-in-2018/Sun, 22 Jul 2018 08:27:43 +0000https://yanirseroussi.com/2018/07/22/defining-data-science-in-2018/Updating my definition of data science to match changes in the field. It is now broader than before, but its ultimate goal is still to support decisions.Customer lifetime value and the proliferation of misinformation on the internethttps://yanirseroussi.com/2017/01/08/customer-lifetime-value-and-the-proliferation-of-misinformation-on-the-internet/Sun, 08 Jan 2017 20:02:30 +0000https://yanirseroussi.com/2017/01/08/customer-lifetime-value-and-the-proliferation-of-misinformation-on-the-internet/There&rsquo;s a lot of misleading content on the estimation of customer lifetime value. Here&rsquo;s what I learned about doing it well.If you don’t pay attention, data can drive you off a cliffhttps://yanirseroussi.com/2016/08/21/seven-ways-to-be-data-driven-off-a-cliff/Sun, 21 Aug 2016 21:34:17 +0000https://yanirseroussi.com/2016/08/21/seven-ways-to-be-data-driven-off-a-cliff/Seven common mistakes to avoid when working with data, such as ignoring uncertainty and confusing observed and unobserved quantities.Making Bayesian A/B testing more accessiblehttps://yanirseroussi.com/2016/06/19/making-bayesian-ab-testing-more-accessible/Sun, 19 Jun 2016 10:32:15 +0000https://yanirseroussi.com/2016/06/19/making-bayesian-ab-testing-more-accessible/A web tool I built to interpret A/B test results in a Bayesian way, including prior specification, visualisations, and decision rules.Why you should stop worrying about deep learning and deepen your understanding of causality insteadhttps://yanirseroussi.com/2016/02/14/why-you-should-stop-worrying-about-deep-learning-and-deepen-your-understanding-of-causality-instead/Sun, 14 Feb 2016 11:04:11 +0000https://yanirseroussi.com/2016/02/14/why-you-should-stop-worrying-about-deep-learning-and-deepen-your-understanding-of-causality-instead/Causality is often overlooked but is of much higher relevance to most data scientists than deep learning.This holiday season, give me real insightshttps://yanirseroussi.com/2015/12/08/this-holiday-season-give-me-real-insights/Tue, 08 Dec 2015 06:57:25 +0000https://yanirseroussi.com/2015/12/08/this-holiday-season-give-me-real-insights/Some companies present raw data or information as &ldquo;insights&rdquo;. This post surveys some examples, and discusses how they can be turned into real insights. \ No newline at end of file +Analytics on Yanir Seroussi | Data & AI for Startup Impacthttps://yanirseroussi.com/tags/analytics/Recent content in Analytics on Yanir Seroussi | Data & AI for Startup ImpactHugo -- gohugo.ioen-auText and figures licensed under [CC BY-NC-ND 4.0](https://creativecommons.org/licenses/by-nc-nd/4.0/) by [Yanir Seroussi](https://yanirseroussi.com/about/), except where noted otherwiseMon, 10 Jun 2024 04:00:00 +0000Startup data health starts with healthy event trackinghttps://yanirseroussi.com/2024/06/10/startup-data-health-starts-with-healthy-event-tracking/Mon, 10 Jun 2024 04:00:00 +0000https://yanirseroussi.com/2024/06/10/startup-data-health-starts-with-healthy-event-tracking/Expanding on the startup health check question of tracking Kukuyeva&rsquo;s five business aspects as wide events.Assessing a startup's data-to-AI healthhttps://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/Mon, 22 Apr 2024 06:00:00 +0000https://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/Reviewing the areas that should be assessed to determine a startup&rsquo;s opportunities and challenges on the data/AI/ML front.Substance over titles: Your first data hire may be a data scientisthttps://yanirseroussi.com/2024/02/05/substance-over-titles-your-first-data-hire-may-be-a-data-scientist/Mon, 05 Feb 2024 02:45:00 +0000https://yanirseroussi.com/2024/02/05/substance-over-titles-your-first-data-hire-may-be-a-data-scientist/Advice for hiring a startup&rsquo;s first data person: match skills to business needs, consider contractors, and get help from data people.Bootstrapping the right way?https://yanirseroussi.com/2019/10/06/bootstrapping-the-right-way/Sun, 06 Oct 2019 06:48:07 +0000https://yanirseroussi.com/2019/10/06/bootstrapping-the-right-way/Video and summary of a talk I gave at YOW! Data on bootstrap estimation of confidence intervals.Defining data science in 2018https://yanirseroussi.com/2018/07/22/defining-data-science-in-2018/Sun, 22 Jul 2018 08:27:43 +0000https://yanirseroussi.com/2018/07/22/defining-data-science-in-2018/Updating my definition of data science to match changes in the field. It is now broader than before, but its ultimate goal is still to support decisions.Customer lifetime value and the proliferation of misinformation on the internethttps://yanirseroussi.com/2017/01/08/customer-lifetime-value-and-the-proliferation-of-misinformation-on-the-internet/Sun, 08 Jan 2017 20:02:30 +0000https://yanirseroussi.com/2017/01/08/customer-lifetime-value-and-the-proliferation-of-misinformation-on-the-internet/There&rsquo;s a lot of misleading content on the estimation of customer lifetime value. Here&rsquo;s what I learned about doing it well.If you don’t pay attention, data can drive you off a cliffhttps://yanirseroussi.com/2016/08/21/seven-ways-to-be-data-driven-off-a-cliff/Sun, 21 Aug 2016 21:34:17 +0000https://yanirseroussi.com/2016/08/21/seven-ways-to-be-data-driven-off-a-cliff/Seven common mistakes to avoid when working with data, such as ignoring uncertainty and confusing observed and unobserved quantities.Making Bayesian A/B testing more accessiblehttps://yanirseroussi.com/2016/06/19/making-bayesian-ab-testing-more-accessible/Sun, 19 Jun 2016 10:32:15 +0000https://yanirseroussi.com/2016/06/19/making-bayesian-ab-testing-more-accessible/A web tool I built to interpret A/B test results in a Bayesian way, including prior specification, visualisations, and decision rules.Why you should stop worrying about deep learning and deepen your understanding of causality insteadhttps://yanirseroussi.com/2016/02/14/why-you-should-stop-worrying-about-deep-learning-and-deepen-your-understanding-of-causality-instead/Sun, 14 Feb 2016 11:04:11 +0000https://yanirseroussi.com/2016/02/14/why-you-should-stop-worrying-about-deep-learning-and-deepen-your-understanding-of-causality-instead/Causality is often overlooked but is of much higher relevance to most data scientists than deep learning.This holiday season, give me real insightshttps://yanirseroussi.com/2015/12/08/this-holiday-season-give-me-real-insights/Tue, 08 Dec 2015 06:57:25 +0000https://yanirseroussi.com/2015/12/08/this-holiday-season-give-me-real-insights/Some companies present raw data or information as &ldquo;insights&rdquo;. This post surveys some examples, and discusses how they can be turned into real insights. \ No newline at end of file diff --git a/tags/business/index.html b/tags/business/index.html index aabd4bc56..bf9179918 100644 --- a/tags/business/index.html +++ b/tags/business/index.html @@ -2,7 +2,7 @@

      Business

      How to avoid startups with poor development processes

      Questions that prospective data specialists and engineers should ask about development processes before accepting a startup role.

      June 3, 2024

      Plumbing, Decisions, and Automation: De-hyping Data & AI

      Three essential questions to understand where an organisation stands when it comes to Data & AI (with zero hype).

      May 27, 2024

      Adapting to the economy of algorithms

      Overview of the book The Economy of Algorithms by Marek Kowalkiewicz.

      May 25, 2024

      Question startup culture before accepting a data-to-AI role

      Eight questions that prospective data-to-AI employees should ask about a startup’s work and data culture.

      May 20, 2024

      Probing the People aspects of an early-stage startup

      Ten questions that prospective employees should ask about a startup’s team, especially for data-centric roles.

      May 13, 2024

      Business questions to ask before taking a startup data role

      Fourteen questions that prospective employees should ask about a startup’s business model and product, especially for data-focused roles.

      May 6, 2024

      Mentorship and the art of actionable advice

      Reflections on what it takes to package expertise and deliver timely, actionable advice outside the context of employee relationships.

      April 29, 2024

      Assessing a startup's data-to-AI health

      Reviewing the areas that should be assessed to determine a startup’s opportunities and challenges on the data/AI/ML front.

      April 22, 2024

      LinkedIn is a teachable skill

      An high-level overview of things I learned from Justin Welsh’s LinkedIn Operating System course.

      April 11, 2024

      The three Cs of indie consulting: Confidence, Cash, and Connections

      Jonathan Stark makes a compelling argument why you should have the three Cs before quitting your job to go solo consulting.

      February 17, 2024

      Substance over titles: Your first data hire may be a data scientist

      Advice for hiring a startup’s first data person: match skills to business needs, consider contractors, and get help from data people.

      February 5, 2024

      Psychographic specialisations may work for discipline generalists

      When focusing on a market segment defined by personal beliefs, it’s often fine to position yourself as a generalist in your craft.

      January 9, 2024

      The power of parasocial relationships

      Repeated exposure to media personas creates relationships that help justify premium fees.

      January 8, 2024

      Positioning is a common problem for data scientists

      With the commodification of data scientists, the problem of positioning has become more common: My takeaways from Genevieve Hayes interviewing Jonathan Stark.

      December 18, 2023

      The lines between solo consulting and product building are blurry

      It turns out that problems like finding a niche and defining the ideal clients are key to any solo business.

      September 25, 2023

      The Minimalist Entrepreneur is too prescriptive for me

      While I found the story of Gumroad interesting, The Minimalist Entrepreneur seems to over-generalise from the founder’s experience.

      August 21, 2023

      Revisiting Start Small, Stay Small in 2023 (Chapter 2)

      A summary of the second chapter of Rob Walling’s Start Small, Stay Small, along with my thoughts & reflections.

      August 17, 2023

      Revisiting Start Small, Stay Small in 2023 (Chapter 1)

      A summary of the first chapter of Rob Walling’s Start Small, Stay Small, along with my thoughts & reflections.

      August 16, 2023

      Software commodities are eating interesting data science work

      Being a data scientist can sometimes feel like a race against software commodities that replace interesting work. What can one do to remain relevant?

      January 11, 2020

      Defining data science in 2018

      Updating my definition of data science to match changes in the field. It is now broader than before, but its ultimate goal is still to support decisions.

      July 22, 2018

      Customer lifetime value and the proliferation of misinformation on the internet

      There’s a lot of misleading content on the estimation of customer lifetime value. Here’s what I learned about doing it well.

      January 8, 2017

      If you don’t pay attention, data can drive you off a cliff

      Seven common mistakes to avoid when working with data, such as ignoring uncertainty and confusing observed and unobserved quantities.

      August 21, 2016

      Is Data Scientist a useless job title?

      It seems like anyone who touches data can call themselves a data scientist, which makes the title useless. The work they do can still be useful, though.

      August 4, 2016

      You don’t need a data scientist (yet)

      Hiring data scientists prematurely is wasteful and frustrating. Here are some questions to ask before you hire your first data scientist.

      August 24, 2015

      The long road to a lifestyle business

      Progress since leaving my last full-time job and setting on an independent path that includes data science consulting and work on my own projects.

      March 22, 2015

      BCRecommender Traction Update

      Update on BCRecommender traction using three channels: blogger outreach, search engine optimisation, and content marketing.

      November 5, 2014

      Applying the Traction Book’s Bullseye framework to BCRecommender

      Ranking 19 channels with the goal of getting traction for BCRecommender.

      September 24, 2014

      Data’s hierarchy of needs

      Discussing the hierarchy of needs proposed by Jay Kreps. Key takeaway: Data-driven algorithms & insights can only be as good as the underlying data.

      August 17, 2014
      Text and figures licensed under CC BY-NC-ND 4.0 by Yanir Seroussi, except where noted otherwise  | +">

      Business

      Startup data health starts with healthy event tracking

      Expanding on the startup health check question of tracking Kukuyeva’s five business aspects as wide events.

      June 10, 2024

      How to avoid startups with poor development processes

      Questions that prospective data specialists and engineers should ask about development processes before accepting a startup role.

      June 3, 2024

      Plumbing, Decisions, and Automation: De-hyping Data & AI

      Three essential questions to understand where an organisation stands when it comes to Data & AI (with zero hype).

      May 27, 2024

      Adapting to the economy of algorithms

      Overview of the book The Economy of Algorithms by Marek Kowalkiewicz.

      May 25, 2024

      Question startup culture before accepting a data-to-AI role

      Eight questions that prospective data-to-AI employees should ask about a startup’s work and data culture.

      May 20, 2024

      Probing the People aspects of an early-stage startup

      Ten questions that prospective employees should ask about a startup’s team, especially for data-centric roles.

      May 13, 2024

      Business questions to ask before taking a startup data role

      Fourteen questions that prospective employees should ask about a startup’s business model and product, especially for data-focused roles.

      May 6, 2024

      Mentorship and the art of actionable advice

      Reflections on what it takes to package expertise and deliver timely, actionable advice outside the context of employee relationships.

      April 29, 2024

      Assessing a startup's data-to-AI health

      Reviewing the areas that should be assessed to determine a startup’s opportunities and challenges on the data/AI/ML front.

      April 22, 2024

      LinkedIn is a teachable skill

      An high-level overview of things I learned from Justin Welsh’s LinkedIn Operating System course.

      April 11, 2024

      The three Cs of indie consulting: Confidence, Cash, and Connections

      Jonathan Stark makes a compelling argument why you should have the three Cs before quitting your job to go solo consulting.

      February 17, 2024

      Substance over titles: Your first data hire may be a data scientist

      Advice for hiring a startup’s first data person: match skills to business needs, consider contractors, and get help from data people.

      February 5, 2024

      Psychographic specialisations may work for discipline generalists

      When focusing on a market segment defined by personal beliefs, it’s often fine to position yourself as a generalist in your craft.

      January 9, 2024

      The power of parasocial relationships

      Repeated exposure to media personas creates relationships that help justify premium fees.

      January 8, 2024

      Positioning is a common problem for data scientists

      With the commodification of data scientists, the problem of positioning has become more common: My takeaways from Genevieve Hayes interviewing Jonathan Stark.

      December 18, 2023

      The lines between solo consulting and product building are blurry

      It turns out that problems like finding a niche and defining the ideal clients are key to any solo business.

      September 25, 2023

      The Minimalist Entrepreneur is too prescriptive for me

      While I found the story of Gumroad interesting, The Minimalist Entrepreneur seems to over-generalise from the founder’s experience.

      August 21, 2023

      Revisiting Start Small, Stay Small in 2023 (Chapter 2)

      A summary of the second chapter of Rob Walling’s Start Small, Stay Small, along with my thoughts & reflections.

      August 17, 2023

      Revisiting Start Small, Stay Small in 2023 (Chapter 1)

      A summary of the first chapter of Rob Walling’s Start Small, Stay Small, along with my thoughts & reflections.

      August 16, 2023

      Software commodities are eating interesting data science work

      Being a data scientist can sometimes feel like a race against software commodities that replace interesting work. What can one do to remain relevant?

      January 11, 2020

      Defining data science in 2018

      Updating my definition of data science to match changes in the field. It is now broader than before, but its ultimate goal is still to support decisions.

      July 22, 2018

      Customer lifetime value and the proliferation of misinformation on the internet

      There’s a lot of misleading content on the estimation of customer lifetime value. Here’s what I learned about doing it well.

      January 8, 2017

      If you don’t pay attention, data can drive you off a cliff

      Seven common mistakes to avoid when working with data, such as ignoring uncertainty and confusing observed and unobserved quantities.

      August 21, 2016

      Is Data Scientist a useless job title?

      It seems like anyone who touches data can call themselves a data scientist, which makes the title useless. The work they do can still be useful, though.

      August 4, 2016

      You don’t need a data scientist (yet)

      Hiring data scientists prematurely is wasteful and frustrating. Here are some questions to ask before you hire your first data scientist.

      August 24, 2015

      The long road to a lifestyle business

      Progress since leaving my last full-time job and setting on an independent path that includes data science consulting and work on my own projects.

      March 22, 2015

      BCRecommender Traction Update

      Update on BCRecommender traction using three channels: blogger outreach, search engine optimisation, and content marketing.

      November 5, 2014

      Applying the Traction Book’s Bullseye framework to BCRecommender

      Ranking 19 channels with the goal of getting traction for BCRecommender.

      September 24, 2014

      Data’s hierarchy of needs

      Discussing the hierarchy of needs proposed by Jay Kreps. Key takeaway: Data-driven algorithms & insights can only be as good as the underlying data.

      August 17, 2014
      Text and figures licensed under CC BY-NC-ND 4.0 by Yanir Seroussi, except where noted otherwise  | Powered by Hugo & PaperMod
      \ No newline at end of file diff --git a/tags/business/index.xml b/tags/business/index.xml index 44419ac90..abd03f9f8 100644 --- a/tags/business/index.xml +++ b/tags/business/index.xml @@ -1 +1 @@ -Business on Yanir Seroussi | Data & AI for Startup Impacthttps://yanirseroussi.com/tags/business/Recent content in Business on Yanir Seroussi | Data & AI for Startup ImpactHugo -- gohugo.ioen-auText and figures licensed under [CC BY-NC-ND 4.0](https://creativecommons.org/licenses/by-nc-nd/4.0/) by [Yanir Seroussi](https://yanirseroussi.com/about/), except where noted otherwiseMon, 03 Jun 2024 02:45:00 +0000How to avoid startups with poor development processeshttps://yanirseroussi.com/2024/06/03/how-to-avoid-startups-with-poor-development-processes/Mon, 03 Jun 2024 02:45:00 +0000https://yanirseroussi.com/2024/06/03/how-to-avoid-startups-with-poor-development-processes/Questions that prospective data specialists and engineers should ask about development processes before accepting a startup role.Plumbing, Decisions, and Automation: De-hyping Data & AIhttps://yanirseroussi.com/2024/05/27/plumbing-decisions-and-automation-de-hyping-data-and-ai/Mon, 27 May 2024 02:00:00 +0000https://yanirseroussi.com/2024/05/27/plumbing-decisions-and-automation-de-hyping-data-and-ai/Three essential questions to understand where an organisation stands when it comes to Data &amp; AI (with zero hype).Adapting to the economy of algorithmshttps://yanirseroussi.com/til/2024/05/25/adapting-to-the-economy-of-algorithms/Sat, 25 May 2024 00:00:00 +0000https://yanirseroussi.com/til/2024/05/25/adapting-to-the-economy-of-algorithms/Overview of the book The Economy of Algorithms by Marek Kowalkiewicz.Question startup culture before accepting a data-to-AI rolehttps://yanirseroussi.com/2024/05/20/question-startup-culture-before-accepting-a-data-to-ai-role/Mon, 20 May 2024 02:25:00 +0000https://yanirseroussi.com/2024/05/20/question-startup-culture-before-accepting-a-data-to-ai-role/Eight questions that prospective data-to-AI employees should ask about a startup&rsquo;s work and data culture.Probing the People aspects of an early-stage startuphttps://yanirseroussi.com/2024/05/13/probing-the-people-aspects-of-an-early-stage-startup/Mon, 13 May 2024 02:00:00 +0000https://yanirseroussi.com/2024/05/13/probing-the-people-aspects-of-an-early-stage-startup/Ten questions that prospective employees should ask about a startup&rsquo;s team, especially for data-centric roles.Business questions to ask before taking a startup data rolehttps://yanirseroussi.com/2024/05/06/business-questions-to-ask-before-taking-a-startup-data-role/Mon, 06 May 2024 04:30:00 +0000https://yanirseroussi.com/2024/05/06/business-questions-to-ask-before-taking-a-startup-data-role/Fourteen questions that prospective employees should ask about a startup&rsquo;s business model and product, especially for data-focused roles.Mentorship and the art of actionable advicehttps://yanirseroussi.com/2024/04/29/mentorship-and-the-art-of-actionable-advice/Mon, 29 Apr 2024 06:30:00 +0000https://yanirseroussi.com/2024/04/29/mentorship-and-the-art-of-actionable-advice/Reflections on what it takes to package expertise and deliver timely, actionable advice outside the context of employee relationships.Assessing a startup's data-to-AI healthhttps://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/Mon, 22 Apr 2024 06:00:00 +0000https://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/Reviewing the areas that should be assessed to determine a startup&rsquo;s opportunities and challenges on the data/AI/ML front.LinkedIn is a teachable skillhttps://yanirseroussi.com/til/2024/04/11/linkedin-is-a-teachable-skill/Thu, 11 Apr 2024 01:45:25 +0000https://yanirseroussi.com/til/2024/04/11/linkedin-is-a-teachable-skill/An high-level overview of things I learned from Justin Welsh&rsquo;s LinkedIn Operating System course.The three Cs of indie consulting: Confidence, Cash, and Connectionshttps://yanirseroussi.com/til/2024/02/17/the-three-cs-of-indie-consulting-confidence-cash-and-connections/Sat, 17 Feb 2024 02:00:00 +0000https://yanirseroussi.com/til/2024/02/17/the-three-cs-of-indie-consulting-confidence-cash-and-connections/Jonathan Stark makes a compelling argument why you should have the three Cs before quitting your job to go solo consulting.Substance over titles: Your first data hire may be a data scientisthttps://yanirseroussi.com/2024/02/05/substance-over-titles-your-first-data-hire-may-be-a-data-scientist/Mon, 05 Feb 2024 02:45:00 +0000https://yanirseroussi.com/2024/02/05/substance-over-titles-your-first-data-hire-may-be-a-data-scientist/Advice for hiring a startup&rsquo;s first data person: match skills to business needs, consider contractors, and get help from data people.Psychographic specialisations may work for discipline generalistshttps://yanirseroussi.com/til/2024/01/09/psychographic-specialisations-may-work-for-discipline-generalists/Tue, 09 Jan 2024 03:00:00 +0000https://yanirseroussi.com/til/2024/01/09/psychographic-specialisations-may-work-for-discipline-generalists/When focusing on a market segment defined by personal beliefs, it&rsquo;s often fine to position yourself as a generalist in your craft.The power of parasocial relationshipshttps://yanirseroussi.com/til/2024/01/08/the-power-of-parasocial-relationships/Mon, 08 Jan 2024 06:00:00 +0000https://yanirseroussi.com/til/2024/01/08/the-power-of-parasocial-relationships/Repeated exposure to media personas creates relationships that help justify premium fees.Positioning is a common problem for data scientistshttps://yanirseroussi.com/til/2023/12/18/positioning-is-a-common-problem-for-data-scientists/Mon, 18 Dec 2023 00:30:00 +0000https://yanirseroussi.com/til/2023/12/18/positioning-is-a-common-problem-for-data-scientists/With the commodification of data scientists, the problem of positioning has become more common: My takeaways from Genevieve Hayes interviewing Jonathan Stark.The lines between solo consulting and product building are blurryhttps://yanirseroussi.com/til/2023/09/25/the-lines-between-solo-consulting-and-product-building-are-blurry/Mon, 25 Sep 2023 00:00:00 +0000https://yanirseroussi.com/til/2023/09/25/the-lines-between-solo-consulting-and-product-building-are-blurry/It turns out that problems like finding a niche and defining the ideal clients are key to any solo business.The Minimalist Entrepreneur is too prescriptive for mehttps://yanirseroussi.com/til/2023/08/21/the-minimalist-entrepreneur-is-too-prescriptive-for-me/Mon, 21 Aug 2023 03:15:00 +0000https://yanirseroussi.com/til/2023/08/21/the-minimalist-entrepreneur-is-too-prescriptive-for-me/While I found the story of Gumroad interesting, The Minimalist Entrepreneur seems to over-generalise from the founder&rsquo;s experience.Revisiting Start Small, Stay Small in 2023 (Chapter 2)https://yanirseroussi.com/til/2023/08/17/revisiting-start-small-stay-small-in-2023-chapter-2/Thu, 17 Aug 2023 07:45:00 +0000https://yanirseroussi.com/til/2023/08/17/revisiting-start-small-stay-small-in-2023-chapter-2/A summary of the second chapter of Rob Walling&rsquo;s Start Small, Stay Small, along with my thoughts &amp; reflections.Revisiting Start Small, Stay Small in 2023 (Chapter 1)https://yanirseroussi.com/til/2023/08/16/revisiting-start-small-stay-small-in-2023-chapter-1/Wed, 16 Aug 2023 05:45:00 +0000https://yanirseroussi.com/til/2023/08/16/revisiting-start-small-stay-small-in-2023-chapter-1/A summary of the first chapter of Rob Walling&rsquo;s Start Small, Stay Small, along with my thoughts &amp; reflections.Software commodities are eating interesting data science workhttps://yanirseroussi.com/2020/01/11/software-commodities-are-eating-interesting-data-science-work/Sat, 11 Jan 2020 09:22:35 +0000https://yanirseroussi.com/2020/01/11/software-commodities-are-eating-interesting-data-science-work/Being a data scientist can sometimes feel like a race against software commodities that replace interesting work. What can one do to remain relevant?Defining data science in 2018https://yanirseroussi.com/2018/07/22/defining-data-science-in-2018/Sun, 22 Jul 2018 08:27:43 +0000https://yanirseroussi.com/2018/07/22/defining-data-science-in-2018/Updating my definition of data science to match changes in the field. It is now broader than before, but its ultimate goal is still to support decisions.Customer lifetime value and the proliferation of misinformation on the internethttps://yanirseroussi.com/2017/01/08/customer-lifetime-value-and-the-proliferation-of-misinformation-on-the-internet/Sun, 08 Jan 2017 20:02:30 +0000https://yanirseroussi.com/2017/01/08/customer-lifetime-value-and-the-proliferation-of-misinformation-on-the-internet/There&rsquo;s a lot of misleading content on the estimation of customer lifetime value. Here&rsquo;s what I learned about doing it well.If you don’t pay attention, data can drive you off a cliffhttps://yanirseroussi.com/2016/08/21/seven-ways-to-be-data-driven-off-a-cliff/Sun, 21 Aug 2016 21:34:17 +0000https://yanirseroussi.com/2016/08/21/seven-ways-to-be-data-driven-off-a-cliff/Seven common mistakes to avoid when working with data, such as ignoring uncertainty and confusing observed and unobserved quantities.Is Data Scientist a useless job title?https://yanirseroussi.com/2016/08/04/is-data-scientist-a-useless-job-title/Thu, 04 Aug 2016 22:26:03 +0000https://yanirseroussi.com/2016/08/04/is-data-scientist-a-useless-job-title/It seems like anyone who touches data can call themselves a data scientist, which makes the title useless. The work they do can still be useful, though.You don’t need a data scientist (yet)https://yanirseroussi.com/2015/08/24/you-dont-need-a-data-scientist-yet/Mon, 24 Aug 2015 08:25:30 +0000https://yanirseroussi.com/2015/08/24/you-dont-need-a-data-scientist-yet/Hiring data scientists prematurely is wasteful and frustrating. Here are some questions to ask before you hire your first data scientist.The long road to a lifestyle businesshttps://yanirseroussi.com/2015/03/22/the-long-road-to-a-lifestyle-business/Sun, 22 Mar 2015 09:43:47 +0000https://yanirseroussi.com/2015/03/22/the-long-road-to-a-lifestyle-business/Progress since leaving my last full-time job and setting on an independent path that includes data science consulting and work on my own projects.BCRecommender Traction Updatehttps://yanirseroussi.com/2014/11/05/bcrecommender-traction-update/Wed, 05 Nov 2014 02:29:35 +0000https://yanirseroussi.com/2014/11/05/bcrecommender-traction-update/Update on BCRecommender traction using three channels: blogger outreach, search engine optimisation, and content marketing.Applying the Traction Book’s Bullseye framework to BCRecommenderhttps://yanirseroussi.com/2014/09/24/applying-the-traction-books-bullseye-framework-to-bcrecommender/Wed, 24 Sep 2014 04:57:39 +0000https://yanirseroussi.com/2014/09/24/applying-the-traction-books-bullseye-framework-to-bcrecommender/Ranking 19 channels with the goal of getting traction for BCRecommender.Data’s hierarchy of needshttps://yanirseroussi.com/2014/08/17/datas-hierarchy-of-needs/Sun, 17 Aug 2014 13:09:30 +0000https://yanirseroussi.com/2014/08/17/datas-hierarchy-of-needs/Discussing the hierarchy of needs proposed by Jay Kreps. Key takeaway: Data-driven algorithms &amp; insights can only be as good as the underlying data. \ No newline at end of file +Business on Yanir Seroussi | Data & AI for Startup Impacthttps://yanirseroussi.com/tags/business/Recent content in Business on Yanir Seroussi | Data & AI for Startup ImpactHugo -- gohugo.ioen-auText and figures licensed under [CC BY-NC-ND 4.0](https://creativecommons.org/licenses/by-nc-nd/4.0/) by [Yanir Seroussi](https://yanirseroussi.com/about/), except where noted otherwiseMon, 10 Jun 2024 04:00:00 +0000Startup data health starts with healthy event trackinghttps://yanirseroussi.com/2024/06/10/startup-data-health-starts-with-healthy-event-tracking/Mon, 10 Jun 2024 04:00:00 +0000https://yanirseroussi.com/2024/06/10/startup-data-health-starts-with-healthy-event-tracking/Expanding on the startup health check question of tracking Kukuyeva&rsquo;s five business aspects as wide events.How to avoid startups with poor development processeshttps://yanirseroussi.com/2024/06/03/how-to-avoid-startups-with-poor-development-processes/Mon, 03 Jun 2024 02:45:00 +0000https://yanirseroussi.com/2024/06/03/how-to-avoid-startups-with-poor-development-processes/Questions that prospective data specialists and engineers should ask about development processes before accepting a startup role.Plumbing, Decisions, and Automation: De-hyping Data & AIhttps://yanirseroussi.com/2024/05/27/plumbing-decisions-and-automation-de-hyping-data-and-ai/Mon, 27 May 2024 02:00:00 +0000https://yanirseroussi.com/2024/05/27/plumbing-decisions-and-automation-de-hyping-data-and-ai/Three essential questions to understand where an organisation stands when it comes to Data &amp; AI (with zero hype).Adapting to the economy of algorithmshttps://yanirseroussi.com/til/2024/05/25/adapting-to-the-economy-of-algorithms/Sat, 25 May 2024 00:00:00 +0000https://yanirseroussi.com/til/2024/05/25/adapting-to-the-economy-of-algorithms/Overview of the book The Economy of Algorithms by Marek Kowalkiewicz.Question startup culture before accepting a data-to-AI rolehttps://yanirseroussi.com/2024/05/20/question-startup-culture-before-accepting-a-data-to-ai-role/Mon, 20 May 2024 02:25:00 +0000https://yanirseroussi.com/2024/05/20/question-startup-culture-before-accepting-a-data-to-ai-role/Eight questions that prospective data-to-AI employees should ask about a startup&rsquo;s work and data culture.Probing the People aspects of an early-stage startuphttps://yanirseroussi.com/2024/05/13/probing-the-people-aspects-of-an-early-stage-startup/Mon, 13 May 2024 02:00:00 +0000https://yanirseroussi.com/2024/05/13/probing-the-people-aspects-of-an-early-stage-startup/Ten questions that prospective employees should ask about a startup&rsquo;s team, especially for data-centric roles.Business questions to ask before taking a startup data rolehttps://yanirseroussi.com/2024/05/06/business-questions-to-ask-before-taking-a-startup-data-role/Mon, 06 May 2024 04:30:00 +0000https://yanirseroussi.com/2024/05/06/business-questions-to-ask-before-taking-a-startup-data-role/Fourteen questions that prospective employees should ask about a startup&rsquo;s business model and product, especially for data-focused roles.Mentorship and the art of actionable advicehttps://yanirseroussi.com/2024/04/29/mentorship-and-the-art-of-actionable-advice/Mon, 29 Apr 2024 06:30:00 +0000https://yanirseroussi.com/2024/04/29/mentorship-and-the-art-of-actionable-advice/Reflections on what it takes to package expertise and deliver timely, actionable advice outside the context of employee relationships.Assessing a startup's data-to-AI healthhttps://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/Mon, 22 Apr 2024 06:00:00 +0000https://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/Reviewing the areas that should be assessed to determine a startup&rsquo;s opportunities and challenges on the data/AI/ML front.LinkedIn is a teachable skillhttps://yanirseroussi.com/til/2024/04/11/linkedin-is-a-teachable-skill/Thu, 11 Apr 2024 01:45:25 +0000https://yanirseroussi.com/til/2024/04/11/linkedin-is-a-teachable-skill/An high-level overview of things I learned from Justin Welsh&rsquo;s LinkedIn Operating System course.The three Cs of indie consulting: Confidence, Cash, and Connectionshttps://yanirseroussi.com/til/2024/02/17/the-three-cs-of-indie-consulting-confidence-cash-and-connections/Sat, 17 Feb 2024 02:00:00 +0000https://yanirseroussi.com/til/2024/02/17/the-three-cs-of-indie-consulting-confidence-cash-and-connections/Jonathan Stark makes a compelling argument why you should have the three Cs before quitting your job to go solo consulting.Substance over titles: Your first data hire may be a data scientisthttps://yanirseroussi.com/2024/02/05/substance-over-titles-your-first-data-hire-may-be-a-data-scientist/Mon, 05 Feb 2024 02:45:00 +0000https://yanirseroussi.com/2024/02/05/substance-over-titles-your-first-data-hire-may-be-a-data-scientist/Advice for hiring a startup&rsquo;s first data person: match skills to business needs, consider contractors, and get help from data people.Psychographic specialisations may work for discipline generalistshttps://yanirseroussi.com/til/2024/01/09/psychographic-specialisations-may-work-for-discipline-generalists/Tue, 09 Jan 2024 03:00:00 +0000https://yanirseroussi.com/til/2024/01/09/psychographic-specialisations-may-work-for-discipline-generalists/When focusing on a market segment defined by personal beliefs, it&rsquo;s often fine to position yourself as a generalist in your craft.The power of parasocial relationshipshttps://yanirseroussi.com/til/2024/01/08/the-power-of-parasocial-relationships/Mon, 08 Jan 2024 06:00:00 +0000https://yanirseroussi.com/til/2024/01/08/the-power-of-parasocial-relationships/Repeated exposure to media personas creates relationships that help justify premium fees.Positioning is a common problem for data scientistshttps://yanirseroussi.com/til/2023/12/18/positioning-is-a-common-problem-for-data-scientists/Mon, 18 Dec 2023 00:30:00 +0000https://yanirseroussi.com/til/2023/12/18/positioning-is-a-common-problem-for-data-scientists/With the commodification of data scientists, the problem of positioning has become more common: My takeaways from Genevieve Hayes interviewing Jonathan Stark.The lines between solo consulting and product building are blurryhttps://yanirseroussi.com/til/2023/09/25/the-lines-between-solo-consulting-and-product-building-are-blurry/Mon, 25 Sep 2023 00:00:00 +0000https://yanirseroussi.com/til/2023/09/25/the-lines-between-solo-consulting-and-product-building-are-blurry/It turns out that problems like finding a niche and defining the ideal clients are key to any solo business.The Minimalist Entrepreneur is too prescriptive for mehttps://yanirseroussi.com/til/2023/08/21/the-minimalist-entrepreneur-is-too-prescriptive-for-me/Mon, 21 Aug 2023 03:15:00 +0000https://yanirseroussi.com/til/2023/08/21/the-minimalist-entrepreneur-is-too-prescriptive-for-me/While I found the story of Gumroad interesting, The Minimalist Entrepreneur seems to over-generalise from the founder&rsquo;s experience.Revisiting Start Small, Stay Small in 2023 (Chapter 2)https://yanirseroussi.com/til/2023/08/17/revisiting-start-small-stay-small-in-2023-chapter-2/Thu, 17 Aug 2023 07:45:00 +0000https://yanirseroussi.com/til/2023/08/17/revisiting-start-small-stay-small-in-2023-chapter-2/A summary of the second chapter of Rob Walling&rsquo;s Start Small, Stay Small, along with my thoughts &amp; reflections.Revisiting Start Small, Stay Small in 2023 (Chapter 1)https://yanirseroussi.com/til/2023/08/16/revisiting-start-small-stay-small-in-2023-chapter-1/Wed, 16 Aug 2023 05:45:00 +0000https://yanirseroussi.com/til/2023/08/16/revisiting-start-small-stay-small-in-2023-chapter-1/A summary of the first chapter of Rob Walling&rsquo;s Start Small, Stay Small, along with my thoughts &amp; reflections.Software commodities are eating interesting data science workhttps://yanirseroussi.com/2020/01/11/software-commodities-are-eating-interesting-data-science-work/Sat, 11 Jan 2020 09:22:35 +0000https://yanirseroussi.com/2020/01/11/software-commodities-are-eating-interesting-data-science-work/Being a data scientist can sometimes feel like a race against software commodities that replace interesting work. What can one do to remain relevant?Defining data science in 2018https://yanirseroussi.com/2018/07/22/defining-data-science-in-2018/Sun, 22 Jul 2018 08:27:43 +0000https://yanirseroussi.com/2018/07/22/defining-data-science-in-2018/Updating my definition of data science to match changes in the field. It is now broader than before, but its ultimate goal is still to support decisions.Customer lifetime value and the proliferation of misinformation on the internethttps://yanirseroussi.com/2017/01/08/customer-lifetime-value-and-the-proliferation-of-misinformation-on-the-internet/Sun, 08 Jan 2017 20:02:30 +0000https://yanirseroussi.com/2017/01/08/customer-lifetime-value-and-the-proliferation-of-misinformation-on-the-internet/There&rsquo;s a lot of misleading content on the estimation of customer lifetime value. Here&rsquo;s what I learned about doing it well.If you don’t pay attention, data can drive you off a cliffhttps://yanirseroussi.com/2016/08/21/seven-ways-to-be-data-driven-off-a-cliff/Sun, 21 Aug 2016 21:34:17 +0000https://yanirseroussi.com/2016/08/21/seven-ways-to-be-data-driven-off-a-cliff/Seven common mistakes to avoid when working with data, such as ignoring uncertainty and confusing observed and unobserved quantities.Is Data Scientist a useless job title?https://yanirseroussi.com/2016/08/04/is-data-scientist-a-useless-job-title/Thu, 04 Aug 2016 22:26:03 +0000https://yanirseroussi.com/2016/08/04/is-data-scientist-a-useless-job-title/It seems like anyone who touches data can call themselves a data scientist, which makes the title useless. The work they do can still be useful, though.You don’t need a data scientist (yet)https://yanirseroussi.com/2015/08/24/you-dont-need-a-data-scientist-yet/Mon, 24 Aug 2015 08:25:30 +0000https://yanirseroussi.com/2015/08/24/you-dont-need-a-data-scientist-yet/Hiring data scientists prematurely is wasteful and frustrating. Here are some questions to ask before you hire your first data scientist.The long road to a lifestyle businesshttps://yanirseroussi.com/2015/03/22/the-long-road-to-a-lifestyle-business/Sun, 22 Mar 2015 09:43:47 +0000https://yanirseroussi.com/2015/03/22/the-long-road-to-a-lifestyle-business/Progress since leaving my last full-time job and setting on an independent path that includes data science consulting and work on my own projects.BCRecommender Traction Updatehttps://yanirseroussi.com/2014/11/05/bcrecommender-traction-update/Wed, 05 Nov 2014 02:29:35 +0000https://yanirseroussi.com/2014/11/05/bcrecommender-traction-update/Update on BCRecommender traction using three channels: blogger outreach, search engine optimisation, and content marketing.Applying the Traction Book’s Bullseye framework to BCRecommenderhttps://yanirseroussi.com/2014/09/24/applying-the-traction-books-bullseye-framework-to-bcrecommender/Wed, 24 Sep 2014 04:57:39 +0000https://yanirseroussi.com/2014/09/24/applying-the-traction-books-bullseye-framework-to-bcrecommender/Ranking 19 channels with the goal of getting traction for BCRecommender.Data’s hierarchy of needshttps://yanirseroussi.com/2014/08/17/datas-hierarchy-of-needs/Sun, 17 Aug 2014 13:09:30 +0000https://yanirseroussi.com/2014/08/17/datas-hierarchy-of-needs/Discussing the hierarchy of needs proposed by Jay Kreps. Key takeaway: Data-driven algorithms &amp; insights can only be as good as the underlying data. \ No newline at end of file diff --git a/tags/data-science/index.html b/tags/data-science/index.html index f52f81651..22465c392 100644 --- a/tags/data-science/index.html +++ b/tags/data-science/index.html @@ -2,7 +2,7 @@

      Data Science

      Plumbing, Decisions, and Automation: De-hyping Data & AI

      Three essential questions to understand where an organisation stands when it comes to Data & AI (with zero hype).

      May 27, 2024

      Assessing a startup's data-to-AI health

      Reviewing the areas that should be assessed to determine a startup’s opportunities and challenges on the data/AI/ML front.

      April 22, 2024

      My experience as a Data Tech Lead with Work on Climate

      The story of how I joined Work on Climate as a volunteer and became its data tech lead, with lessons applied to consulting & fractional work.

      April 8, 2024

      Questions to consider when using AI for PDF data extraction

      Discussing considerations that arise when attempting to automate the extraction of structured data from PDFs and similar documents.

      March 11, 2024

      Substance over titles: Your first data hire may be a data scientist

      Advice for hiring a startup’s first data person: match skills to business needs, consider contractors, and get help from data people.

      February 5, 2024

      New decade, new tagline: Data & AI for Impact

      Shifting focus to ‘Data & AI for Impact’, with more startup-related content, increased posting frequency, and deeper audience engagement.

      January 19, 2024

      Positioning is a common problem for data scientists

      With the commodification of data scientists, the problem of positioning has become more common: My takeaways from Genevieve Hayes interviewing Jonathan Stark.

      December 18, 2023

      You don't need a proprietary API for static maps

      For many use cases, libraries like cartopy are better than the likes of Mapbox and Google Maps.

      November 21, 2023

      Lessons from reluctant data engineering

      Video and summary of a talk I gave at DataEngBytes Brisbane on what I learned from doing data engineering as part of every data science role I had.

      October 25, 2023

      Google's Rules of Machine Learning still apply in the age of large language models

      Despite the excitement around large language models, building with machine learning remains an engineering problem with established best practices.

      September 21, 2023

      Was data science a failure mode of software engineering?

      Yes, data science projects have suffered from classic software engineering mistakes, but the field is maturing with the rise of new engineering roles.

      June 30, 2023

      Causal Machine Learning is off to a good start, despite some issues

      Reviewing the first three chapters of the book Causal Machine Learning by Robert Osazuwa Ness.

      September 12, 2022

      The mission matters: Moving to climate tech as a data scientist

      Discussing my recent career move into climate tech as a way of doing more to help mitigate dangerous climate change.

      June 6, 2022

      Building useful machine learning tools keeps getting easier: A fish ID case study

      Lessons learned building a fish ID web app with fast.ai and Streamlit, in an attempt to reduce my fear of missing out on the latest deep learning developments.

      March 20, 2022

      Analysis strategies in online A/B experiments: Intention-to-treat, per-protocol, and other lessons from clinical trials

      Epidemiologists analyse clinical trials to estimate the intention-to-treat and per-protocol effects. This post applies their strategies to online experiments.

      January 14, 2022

      Use your human brain to avoid artificial intelligence disasters

      Overview of a talk I gave at a deep learning course, focusing on AI ethics as the need for humans to think on the context and consequences of applying AI.

      November 22, 2021

      My work with Automattic

      Back-dated meta-post that gathers my posts on Automattic blogs into a summary of the work I’ve done with the company.

      October 7, 2021

      Many is not enough: Counting simulations to bootstrap the right way

      Going deeper into correct testing of different methods for bootstrap estimation of confidence intervals.

      August 24, 2020

      Software commodities are eating interesting data science work

      Being a data scientist can sometimes feel like a race against software commodities that replace interesting work. What can one do to remain relevant?

      January 11, 2020

      A day in the life of a remote data scientist

      Video of a talk I gave on remote data science work at the Data Science Sydney meetup.

      December 11, 2019

      Bootstrapping the right way?

      Video and summary of a talk I gave at YOW! Data on bootstrap estimation of confidence intervals.

      October 6, 2019

      Hackers beware: Bootstrap sampling may be harmful

      Bootstrap sampling has been promoted as an easy way of modelling uncertainty to hackers without much statistical knowledge. But things aren’t that simple.

      January 7, 2019

      The most practical causal inference book I’ve read (is still a draft)

      Causal Inference by Miguel Hernán and Jamie Robins is a must-read for anyone interested in the area.

      December 24, 2018

      Reflections on remote data science work

      Discussing the pluses and minuses of remote work eighteen months after joining Automattic as a data scientist.

      November 3, 2018

      Defining data science in 2018

      Updating my definition of data science to match changes in the field. It is now broader than before, but its ultimate goal is still to support decisions.

      July 22, 2018

      Advice for aspiring data scientists and other FAQs

      Frequently asked questions by visitors to this site, especially around entering the data science field.

      October 15, 2017

      My 10-step path to becoming a remote data scientist with Automattic

      I wanted a well-paid data science-y remote job with an established company that offers a good life balance and makes products I care about. I got it eventually.

      July 29, 2017

      Exploring and visualising Reef Life Survey data

      Web tools I built to visualise Reef Life Survey data and assist citizen scientists in underwater visual census work.

      June 3, 2017

      Customer lifetime value and the proliferation of misinformation on the internet

      There’s a lot of misleading content on the estimation of customer lifetime value. Here’s what I learned about doing it well.

      January 8, 2017

      Ask Why! Finding motives, causes, and purpose in data science

      Video and summary of a talk I gave at the Data Science Sydney meetup, about going beyond the what & how of predictive modelling.

      September 19, 2016

      If you don’t pay attention, data can drive you off a cliff

      Seven common mistakes to avoid when working with data, such as ignoring uncertainty and confusing observed and unobserved quantities.

      August 21, 2016

      Is Data Scientist a useless job title?

      It seems like anyone who touches data can call themselves a data scientist, which makes the title useless. The work they do can still be useful, though.

      August 4, 2016

      Making Bayesian A/B testing more accessible

      A web tool I built to interpret A/B test results in a Bayesian way, including prior specification, visualisations, and decision rules.

      June 19, 2016

      Diving deeper into causality: Pearl, Kleinberg, Hill, and untested assumptions

      Discussing the need for untested assumptions and temporality in causal inference. Mostly based on Samantha Kleinberg’s Causality, Probability, and Time.

      May 14, 2016

      The rise of greedy robots

      Is artificial/machine intelligence a future threat? I argue that it’s already here, with greedy robots already dominating our lives.

      March 20, 2016

      Why you should stop worrying about deep learning and deepen your understanding of causality instead

      Causality is often overlooked but is of much higher relevance to most data scientists than deep learning.

      February 14, 2016

      The joys of offline data collection

      Insights on data collection and machine learning from spending a month sailing, diving, and counting fish with Reef Life Survey.

      January 24, 2016

      This holiday season, give me real insights

      Some companies present raw data or information as “insights”. This post surveys some examples, and discusses how they can be turned into real insights.

      December 8, 2015

      The hardest parts of data science

      Defining feasible problems and coming up with reasonable ways of measuring solutions is harder than building accurate models or obtaining clean data.

      November 23, 2015

      Miscommunicating science: Simplistic models, nutritionism, and the art of storytelling

      Nutritionism is a special case of misinterpretation and miscommunication of scientific results – something many data scientists encounter in their work.

      October 19, 2015

      The wonderful world of recommender systems

      Giving an overview of the field and common paradigms, and debunking five common myths about recommender systems.

      October 2, 2015

      You don’t need a data scientist (yet)

      Hiring data scientists prematurely is wasteful and frustrating. Here are some questions to ask before you hire your first data scientist.

      August 24, 2015

      Learning about deep learning through album cover classification

      Progress on my album cover classification project, highlighting lessons that would be useful to others who are getting started with deep learning.

      July 6, 2015

      Hopping on the deep learning bandwagon

      To become proficient at solving data science problems, you need to get your hands dirty. Here, I used album cover classification to learn about deep learning.

      June 6, 2015

      First steps in data science: author-aware sentiment analysis

      I became a data scientist by doing a PhD, but the same steps can be followed without a formal education program.

      May 2, 2015

      My PhD work

      An overview of my PhD in data science / artificial intelligence. Thesis title: Text Mining and Rating Prediction with Topical User Models.

      March 30, 2015

      The long road to a lifestyle business

      Progress since leaving my last full-time job and setting on an independent path that includes data science consulting and work on my own projects.

      March 22, 2015

      Learning to rank for personalised search (Yandex Search Personalisation – Kaggle Competition Summary – Part 2)

      My team’s solution to the Yandex Search Personalisation competition (finished 9th out of 194 teams).

      February 11, 2015

      Is thinking like a search engine possible? (Yandex search personalisation – Kaggle competition summary – part 1)

      Insights on search personalisation and SEO from participating in a Kaggle competition (finished 9th out of 194 teams).

      January 29, 2015

      Stochastic Gradient Boosting: Choosing the Best Number of Iterations

      Exploring an approach to choosing the optimal number of iterations in stochastic gradient boosting, following a bug I found in scikit-learn.

      December 29, 2014

      Fitting noise: Forecasting the sale price of bulldozers (Kaggle competition summary)

      Summary of a Kaggle competition to forecast bulldozer sale price, where I finished 9th out of 476 teams.

      November 19, 2014

      What is data science?

      Data science has been a hot term in the past few years. Still, there isn’t a single definition of the field. This post discusses my favourite definition.

      October 23, 2014

      Greek Media Monitoring Kaggle competition: My approach

      Summary of my approach to the Greek Media Monitoring Kaggle competition, where I finished 6th out of 120 teams.

      October 7, 2014

      Bandcamp recommendation and discovery algorithms

      The recommendation backend for my BCRecommender service for personalised Bandcamp music discovery.

      September 19, 2014

      How to (almost) win Kaggle competitions

      Summary of a talk I gave at the Data Science Sydney meetup with ten tips on almost-winning Kaggle competitions.

      August 24, 2014

      Data’s hierarchy of needs

      Discussing the hierarchy of needs proposed by Jay Kreps. Key takeaway: Data-driven algorithms & insights can only be as good as the underlying data.

      August 17, 2014

      Kaggle competition tips and summaries

      Pointers to all my Kaggle advice posts and competition summaries.

      April 5, 2014

      Kaggle beginner tips

      First post! An email I sent to members of the Data Science Sydney Meetup with tips on how to get started with Kaggle competitions.

      January 19, 2014
      Text and figures licensed under CC BY-NC-ND 4.0 by Yanir Seroussi, except where noted otherwise  | +">

      Data Science

      Startup data health starts with healthy event tracking

      Expanding on the startup health check question of tracking Kukuyeva’s five business aspects as wide events.

      June 10, 2024

      Plumbing, Decisions, and Automation: De-hyping Data & AI

      Three essential questions to understand where an organisation stands when it comes to Data & AI (with zero hype).

      May 27, 2024

      Assessing a startup's data-to-AI health

      Reviewing the areas that should be assessed to determine a startup’s opportunities and challenges on the data/AI/ML front.

      April 22, 2024

      My experience as a Data Tech Lead with Work on Climate

      The story of how I joined Work on Climate as a volunteer and became its data tech lead, with lessons applied to consulting & fractional work.

      April 8, 2024

      Questions to consider when using AI for PDF data extraction

      Discussing considerations that arise when attempting to automate the extraction of structured data from PDFs and similar documents.

      March 11, 2024

      Substance over titles: Your first data hire may be a data scientist

      Advice for hiring a startup’s first data person: match skills to business needs, consider contractors, and get help from data people.

      February 5, 2024

      New decade, new tagline: Data & AI for Impact

      Shifting focus to ‘Data & AI for Impact’, with more startup-related content, increased posting frequency, and deeper audience engagement.

      January 19, 2024

      Positioning is a common problem for data scientists

      With the commodification of data scientists, the problem of positioning has become more common: My takeaways from Genevieve Hayes interviewing Jonathan Stark.

      December 18, 2023

      You don't need a proprietary API for static maps

      For many use cases, libraries like cartopy are better than the likes of Mapbox and Google Maps.

      November 21, 2023

      Lessons from reluctant data engineering

      Video and summary of a talk I gave at DataEngBytes Brisbane on what I learned from doing data engineering as part of every data science role I had.

      October 25, 2023

      Google's Rules of Machine Learning still apply in the age of large language models

      Despite the excitement around large language models, building with machine learning remains an engineering problem with established best practices.

      September 21, 2023

      Was data science a failure mode of software engineering?

      Yes, data science projects have suffered from classic software engineering mistakes, but the field is maturing with the rise of new engineering roles.

      June 30, 2023

      Causal Machine Learning is off to a good start, despite some issues

      Reviewing the first three chapters of the book Causal Machine Learning by Robert Osazuwa Ness.

      September 12, 2022

      The mission matters: Moving to climate tech as a data scientist

      Discussing my recent career move into climate tech as a way of doing more to help mitigate dangerous climate change.

      June 6, 2022

      Building useful machine learning tools keeps getting easier: A fish ID case study

      Lessons learned building a fish ID web app with fast.ai and Streamlit, in an attempt to reduce my fear of missing out on the latest deep learning developments.

      March 20, 2022

      Analysis strategies in online A/B experiments: Intention-to-treat, per-protocol, and other lessons from clinical trials

      Epidemiologists analyse clinical trials to estimate the intention-to-treat and per-protocol effects. This post applies their strategies to online experiments.

      January 14, 2022

      Use your human brain to avoid artificial intelligence disasters

      Overview of a talk I gave at a deep learning course, focusing on AI ethics as the need for humans to think on the context and consequences of applying AI.

      November 22, 2021

      My work with Automattic

      Back-dated meta-post that gathers my posts on Automattic blogs into a summary of the work I’ve done with the company.

      October 7, 2021

      Many is not enough: Counting simulations to bootstrap the right way

      Going deeper into correct testing of different methods for bootstrap estimation of confidence intervals.

      August 24, 2020

      Software commodities are eating interesting data science work

      Being a data scientist can sometimes feel like a race against software commodities that replace interesting work. What can one do to remain relevant?

      January 11, 2020

      A day in the life of a remote data scientist

      Video of a talk I gave on remote data science work at the Data Science Sydney meetup.

      December 11, 2019

      Bootstrapping the right way?

      Video and summary of a talk I gave at YOW! Data on bootstrap estimation of confidence intervals.

      October 6, 2019

      Hackers beware: Bootstrap sampling may be harmful

      Bootstrap sampling has been promoted as an easy way of modelling uncertainty to hackers without much statistical knowledge. But things aren’t that simple.

      January 7, 2019

      The most practical causal inference book I’ve read (is still a draft)

      Causal Inference by Miguel Hernán and Jamie Robins is a must-read for anyone interested in the area.

      December 24, 2018

      Reflections on remote data science work

      Discussing the pluses and minuses of remote work eighteen months after joining Automattic as a data scientist.

      November 3, 2018

      Defining data science in 2018

      Updating my definition of data science to match changes in the field. It is now broader than before, but its ultimate goal is still to support decisions.

      July 22, 2018

      Advice for aspiring data scientists and other FAQs

      Frequently asked questions by visitors to this site, especially around entering the data science field.

      October 15, 2017

      My 10-step path to becoming a remote data scientist with Automattic

      I wanted a well-paid data science-y remote job with an established company that offers a good life balance and makes products I care about. I got it eventually.

      July 29, 2017

      Exploring and visualising Reef Life Survey data

      Web tools I built to visualise Reef Life Survey data and assist citizen scientists in underwater visual census work.

      June 3, 2017

      Customer lifetime value and the proliferation of misinformation on the internet

      There’s a lot of misleading content on the estimation of customer lifetime value. Here’s what I learned about doing it well.

      January 8, 2017

      Ask Why! Finding motives, causes, and purpose in data science

      Video and summary of a talk I gave at the Data Science Sydney meetup, about going beyond the what & how of predictive modelling.

      September 19, 2016

      If you don’t pay attention, data can drive you off a cliff

      Seven common mistakes to avoid when working with data, such as ignoring uncertainty and confusing observed and unobserved quantities.

      August 21, 2016

      Is Data Scientist a useless job title?

      It seems like anyone who touches data can call themselves a data scientist, which makes the title useless. The work they do can still be useful, though.

      August 4, 2016

      Making Bayesian A/B testing more accessible

      A web tool I built to interpret A/B test results in a Bayesian way, including prior specification, visualisations, and decision rules.

      June 19, 2016

      Diving deeper into causality: Pearl, Kleinberg, Hill, and untested assumptions

      Discussing the need for untested assumptions and temporality in causal inference. Mostly based on Samantha Kleinberg’s Causality, Probability, and Time.

      May 14, 2016

      The rise of greedy robots

      Is artificial/machine intelligence a future threat? I argue that it’s already here, with greedy robots already dominating our lives.

      March 20, 2016

      Why you should stop worrying about deep learning and deepen your understanding of causality instead

      Causality is often overlooked but is of much higher relevance to most data scientists than deep learning.

      February 14, 2016

      The joys of offline data collection

      Insights on data collection and machine learning from spending a month sailing, diving, and counting fish with Reef Life Survey.

      January 24, 2016

      This holiday season, give me real insights

      Some companies present raw data or information as “insights”. This post surveys some examples, and discusses how they can be turned into real insights.

      December 8, 2015

      The hardest parts of data science

      Defining feasible problems and coming up with reasonable ways of measuring solutions is harder than building accurate models or obtaining clean data.

      November 23, 2015

      Miscommunicating science: Simplistic models, nutritionism, and the art of storytelling

      Nutritionism is a special case of misinterpretation and miscommunication of scientific results – something many data scientists encounter in their work.

      October 19, 2015

      The wonderful world of recommender systems

      Giving an overview of the field and common paradigms, and debunking five common myths about recommender systems.

      October 2, 2015

      You don’t need a data scientist (yet)

      Hiring data scientists prematurely is wasteful and frustrating. Here are some questions to ask before you hire your first data scientist.

      August 24, 2015

      Learning about deep learning through album cover classification

      Progress on my album cover classification project, highlighting lessons that would be useful to others who are getting started with deep learning.

      July 6, 2015

      Hopping on the deep learning bandwagon

      To become proficient at solving data science problems, you need to get your hands dirty. Here, I used album cover classification to learn about deep learning.

      June 6, 2015

      First steps in data science: author-aware sentiment analysis

      I became a data scientist by doing a PhD, but the same steps can be followed without a formal education program.

      May 2, 2015

      My PhD work

      An overview of my PhD in data science / artificial intelligence. Thesis title: Text Mining and Rating Prediction with Topical User Models.

      March 30, 2015

      The long road to a lifestyle business

      Progress since leaving my last full-time job and setting on an independent path that includes data science consulting and work on my own projects.

      March 22, 2015

      Learning to rank for personalised search (Yandex Search Personalisation – Kaggle Competition Summary – Part 2)

      My team’s solution to the Yandex Search Personalisation competition (finished 9th out of 194 teams).

      February 11, 2015

      Is thinking like a search engine possible? (Yandex search personalisation – Kaggle competition summary – part 1)

      Insights on search personalisation and SEO from participating in a Kaggle competition (finished 9th out of 194 teams).

      January 29, 2015

      Stochastic Gradient Boosting: Choosing the Best Number of Iterations

      Exploring an approach to choosing the optimal number of iterations in stochastic gradient boosting, following a bug I found in scikit-learn.

      December 29, 2014

      Fitting noise: Forecasting the sale price of bulldozers (Kaggle competition summary)

      Summary of a Kaggle competition to forecast bulldozer sale price, where I finished 9th out of 476 teams.

      November 19, 2014

      What is data science?

      Data science has been a hot term in the past few years. Still, there isn’t a single definition of the field. This post discusses my favourite definition.

      October 23, 2014

      Greek Media Monitoring Kaggle competition: My approach

      Summary of my approach to the Greek Media Monitoring Kaggle competition, where I finished 6th out of 120 teams.

      October 7, 2014

      Bandcamp recommendation and discovery algorithms

      The recommendation backend for my BCRecommender service for personalised Bandcamp music discovery.

      September 19, 2014

      How to (almost) win Kaggle competitions

      Summary of a talk I gave at the Data Science Sydney meetup with ten tips on almost-winning Kaggle competitions.

      August 24, 2014

      Data’s hierarchy of needs

      Discussing the hierarchy of needs proposed by Jay Kreps. Key takeaway: Data-driven algorithms & insights can only be as good as the underlying data.

      August 17, 2014

      Kaggle competition tips and summaries

      Pointers to all my Kaggle advice posts and competition summaries.

      April 5, 2014

      Kaggle beginner tips

      First post! An email I sent to members of the Data Science Sydney Meetup with tips on how to get started with Kaggle competitions.

      January 19, 2014
      Text and figures licensed under CC BY-NC-ND 4.0 by Yanir Seroussi, except where noted otherwise  | Powered by Hugo & PaperMod
      \ No newline at end of file diff --git a/tags/data-science/index.xml b/tags/data-science/index.xml index 998a8063f..d5a1fc8aa 100644 --- a/tags/data-science/index.xml +++ b/tags/data-science/index.xml @@ -1 +1 @@ -Data Science on Yanir Seroussi | Data & AI for Startup Impacthttps://yanirseroussi.com/tags/data-science/Recent content in Data Science on Yanir Seroussi | Data & AI for Startup ImpactHugo -- gohugo.ioen-auText and figures licensed under [CC BY-NC-ND 4.0](https://creativecommons.org/licenses/by-nc-nd/4.0/) by [Yanir Seroussi](https://yanirseroussi.com/about/), except where noted otherwiseMon, 27 May 2024 02:00:00 +0000Plumbing, Decisions, and Automation: De-hyping Data & AIhttps://yanirseroussi.com/2024/05/27/plumbing-decisions-and-automation-de-hyping-data-and-ai/Mon, 27 May 2024 02:00:00 +0000https://yanirseroussi.com/2024/05/27/plumbing-decisions-and-automation-de-hyping-data-and-ai/Three essential questions to understand where an organisation stands when it comes to Data &amp; AI (with zero hype).Assessing a startup's data-to-AI healthhttps://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/Mon, 22 Apr 2024 06:00:00 +0000https://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/Reviewing the areas that should be assessed to determine a startup&rsquo;s opportunities and challenges on the data/AI/ML front.My experience as a Data Tech Lead with Work on Climatehttps://yanirseroussi.com/2024/04/08/my-experience-as-a-data-tech-lead-with-work-on-climate/Mon, 08 Apr 2024 02:00:00 +0000https://yanirseroussi.com/2024/04/08/my-experience-as-a-data-tech-lead-with-work-on-climate/The story of how I joined Work on Climate as a volunteer and became its data tech lead, with lessons applied to consulting &amp; fractional work.Questions to consider when using AI for PDF data extractionhttps://yanirseroussi.com/2024/03/11/questions-to-consider-when-using-ai-for-pdf-data-extraction/Mon, 11 Mar 2024 00:00:00 +0000https://yanirseroussi.com/2024/03/11/questions-to-consider-when-using-ai-for-pdf-data-extraction/Discussing considerations that arise when attempting to automate the extraction of structured data from PDFs and similar documents.Substance over titles: Your first data hire may be a data scientisthttps://yanirseroussi.com/2024/02/05/substance-over-titles-your-first-data-hire-may-be-a-data-scientist/Mon, 05 Feb 2024 02:45:00 +0000https://yanirseroussi.com/2024/02/05/substance-over-titles-your-first-data-hire-may-be-a-data-scientist/Advice for hiring a startup&rsquo;s first data person: match skills to business needs, consider contractors, and get help from data people.New decade, new tagline: Data & AI for Impacthttps://yanirseroussi.com/2024/01/19/new-decade-new-tagline-data-and-ai-for-impact/Fri, 19 Jan 2024 00:00:00 +0000https://yanirseroussi.com/2024/01/19/new-decade-new-tagline-data-and-ai-for-impact/Shifting focus to &lsquo;Data &amp; AI for Impact&rsquo;, with more startup-related content, increased posting frequency, and deeper audience engagement.Positioning is a common problem for data scientistshttps://yanirseroussi.com/til/2023/12/18/positioning-is-a-common-problem-for-data-scientists/Mon, 18 Dec 2023 00:30:00 +0000https://yanirseroussi.com/til/2023/12/18/positioning-is-a-common-problem-for-data-scientists/With the commodification of data scientists, the problem of positioning has become more common: My takeaways from Genevieve Hayes interviewing Jonathan Stark.You don't need a proprietary API for static mapshttps://yanirseroussi.com/til/2023/11/21/you-dont-need-a-proprietary-api-for-static-maps/Tue, 21 Nov 2023 06:00:00 +0000https://yanirseroussi.com/til/2023/11/21/you-dont-need-a-proprietary-api-for-static-maps/For many use cases, libraries like cartopy are better than the likes of Mapbox and Google Maps.Lessons from reluctant data engineeringhttps://yanirseroussi.com/2023/10/25/lessons-from-reluctant-data-engineering/Wed, 25 Oct 2023 04:45:00 +0000https://yanirseroussi.com/2023/10/25/lessons-from-reluctant-data-engineering/Video and summary of a talk I gave at DataEngBytes Brisbane on what I learned from doing data engineering as part of every data science role I had.Google's Rules of Machine Learning still apply in the age of large language modelshttps://yanirseroussi.com/til/2023/09/21/googles-rules-of-machine-learning-still-apply-in-the-age-of-large-language-models/Thu, 21 Sep 2023 21:30:00 +0000https://yanirseroussi.com/til/2023/09/21/googles-rules-of-machine-learning-still-apply-in-the-age-of-large-language-models/Despite the excitement around large language models, building with machine learning remains an engineering problem with established best practices.Was data science a failure mode of software engineering?https://yanirseroussi.com/2023/06/30/was-data-science-a-failure-mode-of-software-engineering/Fri, 30 Jun 2023 00:06:30 +0000https://yanirseroussi.com/2023/06/30/was-data-science-a-failure-mode-of-software-engineering/Yes, data science projects have suffered from classic software engineering mistakes, but the field is maturing with the rise of new engineering roles.Causal Machine Learning is off to a good start, despite some issueshttps://yanirseroussi.com/2022/09/12/causal-machine-learning-book-draft-review/Mon, 12 Sep 2022 02:45:00 +0000https://yanirseroussi.com/2022/09/12/causal-machine-learning-book-draft-review/Reviewing the first three chapters of the book Causal Machine Learning by Robert Osazuwa Ness.The mission matters: Moving to climate tech as a data scientisthttps://yanirseroussi.com/2022/06/06/the-mission-matters-moving-to-climate-tech-as-a-data-scientist/Mon, 06 Jun 2022 00:00:00 +0000https://yanirseroussi.com/2022/06/06/the-mission-matters-moving-to-climate-tech-as-a-data-scientist/Discussing my recent career move into climate tech as a way of doing more to help mitigate dangerous climate change.Building useful machine learning tools keeps getting easier: A fish ID case studyhttps://yanirseroussi.com/2022/03/20/building-useful-machine-learning-tools-keeps-getting-easier-a-fish-id-case-study/Sun, 20 Mar 2022 04:30:00 +0000https://yanirseroussi.com/2022/03/20/building-useful-machine-learning-tools-keeps-getting-easier-a-fish-id-case-study/Lessons learned building a fish ID web app with fast.ai and Streamlit, in an attempt to reduce my fear of missing out on the latest deep learning developments.Analysis strategies in online A/B experiments: Intention-to-treat, per-protocol, and other lessons from clinical trialshttps://yanirseroussi.com/2022/01/14/analysis-strategies-in-online-a-b-experiments/Fri, 14 Jan 2022 00:05:40 +0000https://yanirseroussi.com/2022/01/14/analysis-strategies-in-online-a-b-experiments/Epidemiologists analyse clinical trials to estimate the intention-to-treat and per-protocol effects. This post applies their strategies to online experiments.Use your human brain to avoid artificial intelligence disastershttps://yanirseroussi.com/2021/11/22/use-your-human-brain-to-avoid-artificial-intelligence-disasters/Mon, 22 Nov 2021 03:45:00 +0000https://yanirseroussi.com/2021/11/22/use-your-human-brain-to-avoid-artificial-intelligence-disasters/Overview of a talk I gave at a deep learning course, focusing on AI ethics as the need for humans to think on the context and consequences of applying AI.My work with Automattichttps://yanirseroussi.com/2021/10/07/my-work-with-automattic/Thu, 07 Oct 2021 00:00:00 +0000https://yanirseroussi.com/2021/10/07/my-work-with-automattic/Back-dated meta-post that gathers my posts on Automattic blogs into a summary of the work I&rsquo;ve done with the company.Many is not enough: Counting simulations to bootstrap the right wayhttps://yanirseroussi.com/2020/08/24/many-is-not-enough-counting-simulations-to-bootstrap-the-right-way/Mon, 24 Aug 2020 01:35:17 +0000https://yanirseroussi.com/2020/08/24/many-is-not-enough-counting-simulations-to-bootstrap-the-right-way/Going deeper into correct testing of different methods for bootstrap estimation of confidence intervals.Software commodities are eating interesting data science workhttps://yanirseroussi.com/2020/01/11/software-commodities-are-eating-interesting-data-science-work/Sat, 11 Jan 2020 09:22:35 +0000https://yanirseroussi.com/2020/01/11/software-commodities-are-eating-interesting-data-science-work/Being a data scientist can sometimes feel like a race against software commodities that replace interesting work. What can one do to remain relevant?A day in the life of a remote data scientisthttps://yanirseroussi.com/2019/12/12/a-day-in-the-life-of-a-remote-data-scientist/Wed, 11 Dec 2019 22:06:19 +0000https://yanirseroussi.com/2019/12/12/a-day-in-the-life-of-a-remote-data-scientist/Video of a talk I gave on remote data science work at the Data Science Sydney meetup.Bootstrapping the right way?https://yanirseroussi.com/2019/10/06/bootstrapping-the-right-way/Sun, 06 Oct 2019 06:48:07 +0000https://yanirseroussi.com/2019/10/06/bootstrapping-the-right-way/Video and summary of a talk I gave at YOW! Data on bootstrap estimation of confidence intervals.Hackers beware: Bootstrap sampling may be harmfulhttps://yanirseroussi.com/2019/01/08/hackers-beware-bootstrap-sampling-may-be-harmful/Mon, 07 Jan 2019 21:07:56 +0000https://yanirseroussi.com/2019/01/08/hackers-beware-bootstrap-sampling-may-be-harmful/Bootstrap sampling has been promoted as an easy way of modelling uncertainty to hackers without much statistical knowledge. But things aren&rsquo;t that simple.The most practical causal inference book I’ve read (is still a draft)https://yanirseroussi.com/2018/12/24/the-most-practical-causal-inference-book-ive-read-is-still-a-draft/Mon, 24 Dec 2018 02:37:50 +0000https://yanirseroussi.com/2018/12/24/the-most-practical-causal-inference-book-ive-read-is-still-a-draft/Causal Inference by Miguel Hernán and Jamie Robins is a must-read for anyone interested in the area.Reflections on remote data science workhttps://yanirseroussi.com/2018/11/03/reflections-on-remote-data-science-work/Sat, 03 Nov 2018 06:33:13 +0000https://yanirseroussi.com/2018/11/03/reflections-on-remote-data-science-work/Discussing the pluses and minuses of remote work eighteen months after joining Automattic as a data scientist.Defining data science in 2018https://yanirseroussi.com/2018/07/22/defining-data-science-in-2018/Sun, 22 Jul 2018 08:27:43 +0000https://yanirseroussi.com/2018/07/22/defining-data-science-in-2018/Updating my definition of data science to match changes in the field. It is now broader than before, but its ultimate goal is still to support decisions.Advice for aspiring data scientists and other FAQshttps://yanirseroussi.com/2017/10/15/advice-for-aspiring-data-scientists-and-other-faqs/Sun, 15 Oct 2017 09:15:25 +0000https://yanirseroussi.com/2017/10/15/advice-for-aspiring-data-scientists-and-other-faqs/Frequently asked questions by visitors to this site, especially around entering the data science field.My 10-step path to becoming a remote data scientist with Automattichttps://yanirseroussi.com/2017/07/29/my-10-step-path-to-becoming-a-remote-data-scientist-with-automattic/Sat, 29 Jul 2017 05:39:26 +0000https://yanirseroussi.com/2017/07/29/my-10-step-path-to-becoming-a-remote-data-scientist-with-automattic/I wanted a well-paid data science-y remote job with an established company that offers a good life balance and makes products I care about. I got it eventually.Exploring and visualising Reef Life Survey datahttps://yanirseroussi.com/2017/06/03/exploring-and-visualising-reef-life-survey-data/Sat, 03 Jun 2017 00:49:05 +0000https://yanirseroussi.com/2017/06/03/exploring-and-visualising-reef-life-survey-data/Web tools I built to visualise Reef Life Survey data and assist citizen scientists in underwater visual census work.Customer lifetime value and the proliferation of misinformation on the internethttps://yanirseroussi.com/2017/01/08/customer-lifetime-value-and-the-proliferation-of-misinformation-on-the-internet/Sun, 08 Jan 2017 20:02:30 +0000https://yanirseroussi.com/2017/01/08/customer-lifetime-value-and-the-proliferation-of-misinformation-on-the-internet/There&rsquo;s a lot of misleading content on the estimation of customer lifetime value. Here&rsquo;s what I learned about doing it well.Ask Why! Finding motives, causes, and purpose in data sciencehttps://yanirseroussi.com/2016/09/19/ask-why-finding-motives-causes-and-purpose-in-data-science/Mon, 19 Sep 2016 21:28:44 +0000https://yanirseroussi.com/2016/09/19/ask-why-finding-motives-causes-and-purpose-in-data-science/Video and summary of a talk I gave at the Data Science Sydney meetup, about going beyond the what &amp; how of predictive modelling.If you don’t pay attention, data can drive you off a cliffhttps://yanirseroussi.com/2016/08/21/seven-ways-to-be-data-driven-off-a-cliff/Sun, 21 Aug 2016 21:34:17 +0000https://yanirseroussi.com/2016/08/21/seven-ways-to-be-data-driven-off-a-cliff/Seven common mistakes to avoid when working with data, such as ignoring uncertainty and confusing observed and unobserved quantities.Is Data Scientist a useless job title?https://yanirseroussi.com/2016/08/04/is-data-scientist-a-useless-job-title/Thu, 04 Aug 2016 22:26:03 +0000https://yanirseroussi.com/2016/08/04/is-data-scientist-a-useless-job-title/It seems like anyone who touches data can call themselves a data scientist, which makes the title useless. The work they do can still be useful, though.Making Bayesian A/B testing more accessiblehttps://yanirseroussi.com/2016/06/19/making-bayesian-ab-testing-more-accessible/Sun, 19 Jun 2016 10:32:15 +0000https://yanirseroussi.com/2016/06/19/making-bayesian-ab-testing-more-accessible/A web tool I built to interpret A/B test results in a Bayesian way, including prior specification, visualisations, and decision rules.Diving deeper into causality: Pearl, Kleinberg, Hill, and untested assumptionshttps://yanirseroussi.com/2016/05/15/diving-deeper-into-causality-pearl-kleinberg-hill-and-untested-assumptions/Sat, 14 May 2016 19:57:03 +0000https://yanirseroussi.com/2016/05/15/diving-deeper-into-causality-pearl-kleinberg-hill-and-untested-assumptions/Discussing the need for untested assumptions and temporality in causal inference. Mostly based on Samantha Kleinberg&rsquo;s Causality, Probability, and Time.The rise of greedy robotshttps://yanirseroussi.com/2016/03/20/the-rise-of-greedy-robots/Sun, 20 Mar 2016 20:33:43 +0000https://yanirseroussi.com/2016/03/20/the-rise-of-greedy-robots/Is artificial/machine intelligence a future threat? I argue that it&rsquo;s already here, with greedy robots already dominating our lives.Why you should stop worrying about deep learning and deepen your understanding of causality insteadhttps://yanirseroussi.com/2016/02/14/why-you-should-stop-worrying-about-deep-learning-and-deepen-your-understanding-of-causality-instead/Sun, 14 Feb 2016 11:04:11 +0000https://yanirseroussi.com/2016/02/14/why-you-should-stop-worrying-about-deep-learning-and-deepen-your-understanding-of-causality-instead/Causality is often overlooked but is of much higher relevance to most data scientists than deep learning.The joys of offline data collectionhttps://yanirseroussi.com/2016/01/24/the-joys-of-offline-data-collection/Sun, 24 Jan 2016 00:32:25 +0000https://yanirseroussi.com/2016/01/24/the-joys-of-offline-data-collection/Insights on data collection and machine learning from spending a month sailing, diving, and counting fish with Reef Life Survey.This holiday season, give me real insightshttps://yanirseroussi.com/2015/12/08/this-holiday-season-give-me-real-insights/Tue, 08 Dec 2015 06:57:25 +0000https://yanirseroussi.com/2015/12/08/this-holiday-season-give-me-real-insights/Some companies present raw data or information as &ldquo;insights&rdquo;. This post surveys some examples, and discusses how they can be turned into real insights.The hardest parts of data sciencehttps://yanirseroussi.com/2015/11/23/the-hardest-parts-of-data-science/Mon, 23 Nov 2015 04:14:21 +0000https://yanirseroussi.com/2015/11/23/the-hardest-parts-of-data-science/Defining feasible problems and coming up with reasonable ways of measuring solutions is harder than building accurate models or obtaining clean data.Miscommunicating science: Simplistic models, nutritionism, and the art of storytellinghttps://yanirseroussi.com/2015/10/19/nutritionism-and-the-need-for-complex-models-to-explain-complex-phenomena/Mon, 19 Oct 2015 00:02:32 +0000https://yanirseroussi.com/2015/10/19/nutritionism-and-the-need-for-complex-models-to-explain-complex-phenomena/Nutritionism is a special case of misinterpretation and miscommunication of scientific results – something many data scientists encounter in their work.The wonderful world of recommender systemshttps://yanirseroussi.com/2015/10/02/the-wonderful-world-of-recommender-systems/Fri, 02 Oct 2015 05:25:57 +0000https://yanirseroussi.com/2015/10/02/the-wonderful-world-of-recommender-systems/Giving an overview of the field and common paradigms, and debunking five common myths about recommender systems.You don’t need a data scientist (yet)https://yanirseroussi.com/2015/08/24/you-dont-need-a-data-scientist-yet/Mon, 24 Aug 2015 08:25:30 +0000https://yanirseroussi.com/2015/08/24/you-dont-need-a-data-scientist-yet/Hiring data scientists prematurely is wasteful and frustrating. Here are some questions to ask before you hire your first data scientist.Learning about deep learning through album cover classificationhttps://yanirseroussi.com/2015/07/06/learning-about-deep-learning-through-album-cover-classification/Mon, 06 Jul 2015 22:21:42 +0000https://yanirseroussi.com/2015/07/06/learning-about-deep-learning-through-album-cover-classification/Progress on my album cover classification project, highlighting lessons that would be useful to others who are getting started with deep learning.Hopping on the deep learning bandwagonhttps://yanirseroussi.com/2015/06/06/hopping-on-the-deep-learning-bandwagon/Sat, 06 Jun 2015 05:00:22 +0000https://yanirseroussi.com/2015/06/06/hopping-on-the-deep-learning-bandwagon/To become proficient at solving data science problems, you need to get your hands dirty. Here, I used album cover classification to learn about deep learning.First steps in data science: author-aware sentiment analysishttps://yanirseroussi.com/2015/05/02/first-steps-in-data-science-author-aware-sentiment-analysis/Sat, 02 May 2015 08:31:10 +0000https://yanirseroussi.com/2015/05/02/first-steps-in-data-science-author-aware-sentiment-analysis/I became a data scientist by doing a PhD, but the same steps can be followed without a formal education program.My PhD workhttps://yanirseroussi.com/phd-work/Mon, 30 Mar 2015 03:23:33 +0000https://yanirseroussi.com/phd-work/An overview of my PhD in data science / artificial intelligence. Thesis title: Text Mining and Rating Prediction with Topical User Models.The long road to a lifestyle businesshttps://yanirseroussi.com/2015/03/22/the-long-road-to-a-lifestyle-business/Sun, 22 Mar 2015 09:43:47 +0000https://yanirseroussi.com/2015/03/22/the-long-road-to-a-lifestyle-business/Progress since leaving my last full-time job and setting on an independent path that includes data science consulting and work on my own projects.Learning to rank for personalised search (Yandex Search Personalisation – Kaggle Competition Summary – Part 2)https://yanirseroussi.com/2015/02/11/learning-to-rank-for-personalised-search-yandex-search-personalisation-kaggle-competition-summary-part-2/Wed, 11 Feb 2015 06:34:17 +0000https://yanirseroussi.com/2015/02/11/learning-to-rank-for-personalised-search-yandex-search-personalisation-kaggle-competition-summary-part-2/My team&rsquo;s solution to the Yandex Search Personalisation competition (finished 9th out of 194 teams).Is thinking like a search engine possible? (Yandex search personalisation – Kaggle competition summary – part 1)https://yanirseroussi.com/2015/01/29/is-thinking-like-a-search-engine-possible-yandex-search-personalisation-kaggle-competition-summary-part-1/Thu, 29 Jan 2015 10:37:39 +0000https://yanirseroussi.com/2015/01/29/is-thinking-like-a-search-engine-possible-yandex-search-personalisation-kaggle-competition-summary-part-1/Insights on search personalisation and SEO from participating in a Kaggle competition (finished 9th out of 194 teams).Stochastic Gradient Boosting: Choosing the Best Number of Iterationshttps://yanirseroussi.com/2014/12/29/stochastic-gradient-boosting-choosing-the-best-number-of-iterations/Mon, 29 Dec 2014 02:30:06 +0000https://yanirseroussi.com/2014/12/29/stochastic-gradient-boosting-choosing-the-best-number-of-iterations/Exploring an approach to choosing the optimal number of iterations in stochastic gradient boosting, following a bug I found in scikit-learn.Fitting noise: Forecasting the sale price of bulldozers (Kaggle competition summary)https://yanirseroussi.com/2014/11/19/fitting-noise-forecasting-the-sale-price-of-bulldozers-kaggle-competition-summary/Wed, 19 Nov 2014 09:17:34 +0000https://yanirseroussi.com/2014/11/19/fitting-noise-forecasting-the-sale-price-of-bulldozers-kaggle-competition-summary/Summary of a Kaggle competition to forecast bulldozer sale price, where I finished 9th out of 476 teams.What is data science?https://yanirseroussi.com/2014/10/23/what-is-data-science/Thu, 23 Oct 2014 03:22:08 +0000https://yanirseroussi.com/2014/10/23/what-is-data-science/Data science has been a hot term in the past few years. Still, there isn&rsquo;t a single definition of the field. This post discusses my favourite definition.Greek Media Monitoring Kaggle competition: My approachhttps://yanirseroussi.com/2014/10/07/greek-media-monitoring-kaggle-competition-my-approach/Tue, 07 Oct 2014 03:21:35 +0000https://yanirseroussi.com/2014/10/07/greek-media-monitoring-kaggle-competition-my-approach/Summary of my approach to the Greek Media Monitoring Kaggle competition, where I finished 6th out of 120 teams.Bandcamp recommendation and discovery algorithmshttps://yanirseroussi.com/2014/09/19/bandcamp-recommendation-and-discovery-algorithms/Fri, 19 Sep 2014 14:26:55 +0000https://yanirseroussi.com/2014/09/19/bandcamp-recommendation-and-discovery-algorithms/The recommendation backend for my BCRecommender service for personalised Bandcamp music discovery.How to (almost) win Kaggle competitionshttps://yanirseroussi.com/2014/08/24/how-to-almost-win-kaggle-competitions/Sun, 24 Aug 2014 12:40:53 +0000https://yanirseroussi.com/2014/08/24/how-to-almost-win-kaggle-competitions/Summary of a talk I gave at the Data Science Sydney meetup with ten tips on almost-winning Kaggle competitions.Data’s hierarchy of needshttps://yanirseroussi.com/2014/08/17/datas-hierarchy-of-needs/Sun, 17 Aug 2014 13:09:30 +0000https://yanirseroussi.com/2014/08/17/datas-hierarchy-of-needs/Discussing the hierarchy of needs proposed by Jay Kreps. Key takeaway: Data-driven algorithms &amp; insights can only be as good as the underlying data.Kaggle competition tips and summarieshttps://yanirseroussi.com/kaggle/Sat, 05 Apr 2014 23:46:10 +0000https://yanirseroussi.com/kaggle/Pointers to all my Kaggle advice posts and competition summaries.Kaggle beginner tipshttps://yanirseroussi.com/2014/01/19/kaggle-beginner-tips/Sun, 19 Jan 2014 10:34:28 +0000https://yanirseroussi.com/2014/01/19/kaggle-beginner-tips/First post! An email I sent to members of the Data Science Sydney Meetup with tips on how to get started with Kaggle competitions. \ No newline at end of file +Data Science on Yanir Seroussi | Data & AI for Startup Impacthttps://yanirseroussi.com/tags/data-science/Recent content in Data Science on Yanir Seroussi | Data & AI for Startup ImpactHugo -- gohugo.ioen-auText and figures licensed under [CC BY-NC-ND 4.0](https://creativecommons.org/licenses/by-nc-nd/4.0/) by [Yanir Seroussi](https://yanirseroussi.com/about/), except where noted otherwiseMon, 10 Jun 2024 04:00:00 +0000Startup data health starts with healthy event trackinghttps://yanirseroussi.com/2024/06/10/startup-data-health-starts-with-healthy-event-tracking/Mon, 10 Jun 2024 04:00:00 +0000https://yanirseroussi.com/2024/06/10/startup-data-health-starts-with-healthy-event-tracking/Expanding on the startup health check question of tracking Kukuyeva&rsquo;s five business aspects as wide events.Plumbing, Decisions, and Automation: De-hyping Data & AIhttps://yanirseroussi.com/2024/05/27/plumbing-decisions-and-automation-de-hyping-data-and-ai/Mon, 27 May 2024 02:00:00 +0000https://yanirseroussi.com/2024/05/27/plumbing-decisions-and-automation-de-hyping-data-and-ai/Three essential questions to understand where an organisation stands when it comes to Data &amp; AI (with zero hype).Assessing a startup's data-to-AI healthhttps://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/Mon, 22 Apr 2024 06:00:00 +0000https://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/Reviewing the areas that should be assessed to determine a startup&rsquo;s opportunities and challenges on the data/AI/ML front.My experience as a Data Tech Lead with Work on Climatehttps://yanirseroussi.com/2024/04/08/my-experience-as-a-data-tech-lead-with-work-on-climate/Mon, 08 Apr 2024 02:00:00 +0000https://yanirseroussi.com/2024/04/08/my-experience-as-a-data-tech-lead-with-work-on-climate/The story of how I joined Work on Climate as a volunteer and became its data tech lead, with lessons applied to consulting &amp; fractional work.Questions to consider when using AI for PDF data extractionhttps://yanirseroussi.com/2024/03/11/questions-to-consider-when-using-ai-for-pdf-data-extraction/Mon, 11 Mar 2024 00:00:00 +0000https://yanirseroussi.com/2024/03/11/questions-to-consider-when-using-ai-for-pdf-data-extraction/Discussing considerations that arise when attempting to automate the extraction of structured data from PDFs and similar documents.Substance over titles: Your first data hire may be a data scientisthttps://yanirseroussi.com/2024/02/05/substance-over-titles-your-first-data-hire-may-be-a-data-scientist/Mon, 05 Feb 2024 02:45:00 +0000https://yanirseroussi.com/2024/02/05/substance-over-titles-your-first-data-hire-may-be-a-data-scientist/Advice for hiring a startup&rsquo;s first data person: match skills to business needs, consider contractors, and get help from data people.New decade, new tagline: Data & AI for Impacthttps://yanirseroussi.com/2024/01/19/new-decade-new-tagline-data-and-ai-for-impact/Fri, 19 Jan 2024 00:00:00 +0000https://yanirseroussi.com/2024/01/19/new-decade-new-tagline-data-and-ai-for-impact/Shifting focus to &lsquo;Data &amp; AI for Impact&rsquo;, with more startup-related content, increased posting frequency, and deeper audience engagement.Positioning is a common problem for data scientistshttps://yanirseroussi.com/til/2023/12/18/positioning-is-a-common-problem-for-data-scientists/Mon, 18 Dec 2023 00:30:00 +0000https://yanirseroussi.com/til/2023/12/18/positioning-is-a-common-problem-for-data-scientists/With the commodification of data scientists, the problem of positioning has become more common: My takeaways from Genevieve Hayes interviewing Jonathan Stark.You don't need a proprietary API for static mapshttps://yanirseroussi.com/til/2023/11/21/you-dont-need-a-proprietary-api-for-static-maps/Tue, 21 Nov 2023 06:00:00 +0000https://yanirseroussi.com/til/2023/11/21/you-dont-need-a-proprietary-api-for-static-maps/For many use cases, libraries like cartopy are better than the likes of Mapbox and Google Maps.Lessons from reluctant data engineeringhttps://yanirseroussi.com/2023/10/25/lessons-from-reluctant-data-engineering/Wed, 25 Oct 2023 04:45:00 +0000https://yanirseroussi.com/2023/10/25/lessons-from-reluctant-data-engineering/Video and summary of a talk I gave at DataEngBytes Brisbane on what I learned from doing data engineering as part of every data science role I had.Google's Rules of Machine Learning still apply in the age of large language modelshttps://yanirseroussi.com/til/2023/09/21/googles-rules-of-machine-learning-still-apply-in-the-age-of-large-language-models/Thu, 21 Sep 2023 21:30:00 +0000https://yanirseroussi.com/til/2023/09/21/googles-rules-of-machine-learning-still-apply-in-the-age-of-large-language-models/Despite the excitement around large language models, building with machine learning remains an engineering problem with established best practices.Was data science a failure mode of software engineering?https://yanirseroussi.com/2023/06/30/was-data-science-a-failure-mode-of-software-engineering/Fri, 30 Jun 2023 00:06:30 +0000https://yanirseroussi.com/2023/06/30/was-data-science-a-failure-mode-of-software-engineering/Yes, data science projects have suffered from classic software engineering mistakes, but the field is maturing with the rise of new engineering roles.Causal Machine Learning is off to a good start, despite some issueshttps://yanirseroussi.com/2022/09/12/causal-machine-learning-book-draft-review/Mon, 12 Sep 2022 02:45:00 +0000https://yanirseroussi.com/2022/09/12/causal-machine-learning-book-draft-review/Reviewing the first three chapters of the book Causal Machine Learning by Robert Osazuwa Ness.The mission matters: Moving to climate tech as a data scientisthttps://yanirseroussi.com/2022/06/06/the-mission-matters-moving-to-climate-tech-as-a-data-scientist/Mon, 06 Jun 2022 00:00:00 +0000https://yanirseroussi.com/2022/06/06/the-mission-matters-moving-to-climate-tech-as-a-data-scientist/Discussing my recent career move into climate tech as a way of doing more to help mitigate dangerous climate change.Building useful machine learning tools keeps getting easier: A fish ID case studyhttps://yanirseroussi.com/2022/03/20/building-useful-machine-learning-tools-keeps-getting-easier-a-fish-id-case-study/Sun, 20 Mar 2022 04:30:00 +0000https://yanirseroussi.com/2022/03/20/building-useful-machine-learning-tools-keeps-getting-easier-a-fish-id-case-study/Lessons learned building a fish ID web app with fast.ai and Streamlit, in an attempt to reduce my fear of missing out on the latest deep learning developments.Analysis strategies in online A/B experiments: Intention-to-treat, per-protocol, and other lessons from clinical trialshttps://yanirseroussi.com/2022/01/14/analysis-strategies-in-online-a-b-experiments/Fri, 14 Jan 2022 00:05:40 +0000https://yanirseroussi.com/2022/01/14/analysis-strategies-in-online-a-b-experiments/Epidemiologists analyse clinical trials to estimate the intention-to-treat and per-protocol effects. This post applies their strategies to online experiments.Use your human brain to avoid artificial intelligence disastershttps://yanirseroussi.com/2021/11/22/use-your-human-brain-to-avoid-artificial-intelligence-disasters/Mon, 22 Nov 2021 03:45:00 +0000https://yanirseroussi.com/2021/11/22/use-your-human-brain-to-avoid-artificial-intelligence-disasters/Overview of a talk I gave at a deep learning course, focusing on AI ethics as the need for humans to think on the context and consequences of applying AI.My work with Automattichttps://yanirseroussi.com/2021/10/07/my-work-with-automattic/Thu, 07 Oct 2021 00:00:00 +0000https://yanirseroussi.com/2021/10/07/my-work-with-automattic/Back-dated meta-post that gathers my posts on Automattic blogs into a summary of the work I&rsquo;ve done with the company.Many is not enough: Counting simulations to bootstrap the right wayhttps://yanirseroussi.com/2020/08/24/many-is-not-enough-counting-simulations-to-bootstrap-the-right-way/Mon, 24 Aug 2020 01:35:17 +0000https://yanirseroussi.com/2020/08/24/many-is-not-enough-counting-simulations-to-bootstrap-the-right-way/Going deeper into correct testing of different methods for bootstrap estimation of confidence intervals.Software commodities are eating interesting data science workhttps://yanirseroussi.com/2020/01/11/software-commodities-are-eating-interesting-data-science-work/Sat, 11 Jan 2020 09:22:35 +0000https://yanirseroussi.com/2020/01/11/software-commodities-are-eating-interesting-data-science-work/Being a data scientist can sometimes feel like a race against software commodities that replace interesting work. What can one do to remain relevant?A day in the life of a remote data scientisthttps://yanirseroussi.com/2019/12/12/a-day-in-the-life-of-a-remote-data-scientist/Wed, 11 Dec 2019 22:06:19 +0000https://yanirseroussi.com/2019/12/12/a-day-in-the-life-of-a-remote-data-scientist/Video of a talk I gave on remote data science work at the Data Science Sydney meetup.Bootstrapping the right way?https://yanirseroussi.com/2019/10/06/bootstrapping-the-right-way/Sun, 06 Oct 2019 06:48:07 +0000https://yanirseroussi.com/2019/10/06/bootstrapping-the-right-way/Video and summary of a talk I gave at YOW! Data on bootstrap estimation of confidence intervals.Hackers beware: Bootstrap sampling may be harmfulhttps://yanirseroussi.com/2019/01/08/hackers-beware-bootstrap-sampling-may-be-harmful/Mon, 07 Jan 2019 21:07:56 +0000https://yanirseroussi.com/2019/01/08/hackers-beware-bootstrap-sampling-may-be-harmful/Bootstrap sampling has been promoted as an easy way of modelling uncertainty to hackers without much statistical knowledge. But things aren&rsquo;t that simple.The most practical causal inference book I’ve read (is still a draft)https://yanirseroussi.com/2018/12/24/the-most-practical-causal-inference-book-ive-read-is-still-a-draft/Mon, 24 Dec 2018 02:37:50 +0000https://yanirseroussi.com/2018/12/24/the-most-practical-causal-inference-book-ive-read-is-still-a-draft/Causal Inference by Miguel Hernán and Jamie Robins is a must-read for anyone interested in the area.Reflections on remote data science workhttps://yanirseroussi.com/2018/11/03/reflections-on-remote-data-science-work/Sat, 03 Nov 2018 06:33:13 +0000https://yanirseroussi.com/2018/11/03/reflections-on-remote-data-science-work/Discussing the pluses and minuses of remote work eighteen months after joining Automattic as a data scientist.Defining data science in 2018https://yanirseroussi.com/2018/07/22/defining-data-science-in-2018/Sun, 22 Jul 2018 08:27:43 +0000https://yanirseroussi.com/2018/07/22/defining-data-science-in-2018/Updating my definition of data science to match changes in the field. It is now broader than before, but its ultimate goal is still to support decisions.Advice for aspiring data scientists and other FAQshttps://yanirseroussi.com/2017/10/15/advice-for-aspiring-data-scientists-and-other-faqs/Sun, 15 Oct 2017 09:15:25 +0000https://yanirseroussi.com/2017/10/15/advice-for-aspiring-data-scientists-and-other-faqs/Frequently asked questions by visitors to this site, especially around entering the data science field.My 10-step path to becoming a remote data scientist with Automattichttps://yanirseroussi.com/2017/07/29/my-10-step-path-to-becoming-a-remote-data-scientist-with-automattic/Sat, 29 Jul 2017 05:39:26 +0000https://yanirseroussi.com/2017/07/29/my-10-step-path-to-becoming-a-remote-data-scientist-with-automattic/I wanted a well-paid data science-y remote job with an established company that offers a good life balance and makes products I care about. I got it eventually.Exploring and visualising Reef Life Survey datahttps://yanirseroussi.com/2017/06/03/exploring-and-visualising-reef-life-survey-data/Sat, 03 Jun 2017 00:49:05 +0000https://yanirseroussi.com/2017/06/03/exploring-and-visualising-reef-life-survey-data/Web tools I built to visualise Reef Life Survey data and assist citizen scientists in underwater visual census work.Customer lifetime value and the proliferation of misinformation on the internethttps://yanirseroussi.com/2017/01/08/customer-lifetime-value-and-the-proliferation-of-misinformation-on-the-internet/Sun, 08 Jan 2017 20:02:30 +0000https://yanirseroussi.com/2017/01/08/customer-lifetime-value-and-the-proliferation-of-misinformation-on-the-internet/There&rsquo;s a lot of misleading content on the estimation of customer lifetime value. Here&rsquo;s what I learned about doing it well.Ask Why! Finding motives, causes, and purpose in data sciencehttps://yanirseroussi.com/2016/09/19/ask-why-finding-motives-causes-and-purpose-in-data-science/Mon, 19 Sep 2016 21:28:44 +0000https://yanirseroussi.com/2016/09/19/ask-why-finding-motives-causes-and-purpose-in-data-science/Video and summary of a talk I gave at the Data Science Sydney meetup, about going beyond the what &amp; how of predictive modelling.If you don’t pay attention, data can drive you off a cliffhttps://yanirseroussi.com/2016/08/21/seven-ways-to-be-data-driven-off-a-cliff/Sun, 21 Aug 2016 21:34:17 +0000https://yanirseroussi.com/2016/08/21/seven-ways-to-be-data-driven-off-a-cliff/Seven common mistakes to avoid when working with data, such as ignoring uncertainty and confusing observed and unobserved quantities.Is Data Scientist a useless job title?https://yanirseroussi.com/2016/08/04/is-data-scientist-a-useless-job-title/Thu, 04 Aug 2016 22:26:03 +0000https://yanirseroussi.com/2016/08/04/is-data-scientist-a-useless-job-title/It seems like anyone who touches data can call themselves a data scientist, which makes the title useless. The work they do can still be useful, though.Making Bayesian A/B testing more accessiblehttps://yanirseroussi.com/2016/06/19/making-bayesian-ab-testing-more-accessible/Sun, 19 Jun 2016 10:32:15 +0000https://yanirseroussi.com/2016/06/19/making-bayesian-ab-testing-more-accessible/A web tool I built to interpret A/B test results in a Bayesian way, including prior specification, visualisations, and decision rules.Diving deeper into causality: Pearl, Kleinberg, Hill, and untested assumptionshttps://yanirseroussi.com/2016/05/15/diving-deeper-into-causality-pearl-kleinberg-hill-and-untested-assumptions/Sat, 14 May 2016 19:57:03 +0000https://yanirseroussi.com/2016/05/15/diving-deeper-into-causality-pearl-kleinberg-hill-and-untested-assumptions/Discussing the need for untested assumptions and temporality in causal inference. Mostly based on Samantha Kleinberg&rsquo;s Causality, Probability, and Time.The rise of greedy robotshttps://yanirseroussi.com/2016/03/20/the-rise-of-greedy-robots/Sun, 20 Mar 2016 20:33:43 +0000https://yanirseroussi.com/2016/03/20/the-rise-of-greedy-robots/Is artificial/machine intelligence a future threat? I argue that it&rsquo;s already here, with greedy robots already dominating our lives.Why you should stop worrying about deep learning and deepen your understanding of causality insteadhttps://yanirseroussi.com/2016/02/14/why-you-should-stop-worrying-about-deep-learning-and-deepen-your-understanding-of-causality-instead/Sun, 14 Feb 2016 11:04:11 +0000https://yanirseroussi.com/2016/02/14/why-you-should-stop-worrying-about-deep-learning-and-deepen-your-understanding-of-causality-instead/Causality is often overlooked but is of much higher relevance to most data scientists than deep learning.The joys of offline data collectionhttps://yanirseroussi.com/2016/01/24/the-joys-of-offline-data-collection/Sun, 24 Jan 2016 00:32:25 +0000https://yanirseroussi.com/2016/01/24/the-joys-of-offline-data-collection/Insights on data collection and machine learning from spending a month sailing, diving, and counting fish with Reef Life Survey.This holiday season, give me real insightshttps://yanirseroussi.com/2015/12/08/this-holiday-season-give-me-real-insights/Tue, 08 Dec 2015 06:57:25 +0000https://yanirseroussi.com/2015/12/08/this-holiday-season-give-me-real-insights/Some companies present raw data or information as &ldquo;insights&rdquo;. This post surveys some examples, and discusses how they can be turned into real insights.The hardest parts of data sciencehttps://yanirseroussi.com/2015/11/23/the-hardest-parts-of-data-science/Mon, 23 Nov 2015 04:14:21 +0000https://yanirseroussi.com/2015/11/23/the-hardest-parts-of-data-science/Defining feasible problems and coming up with reasonable ways of measuring solutions is harder than building accurate models or obtaining clean data.Miscommunicating science: Simplistic models, nutritionism, and the art of storytellinghttps://yanirseroussi.com/2015/10/19/nutritionism-and-the-need-for-complex-models-to-explain-complex-phenomena/Mon, 19 Oct 2015 00:02:32 +0000https://yanirseroussi.com/2015/10/19/nutritionism-and-the-need-for-complex-models-to-explain-complex-phenomena/Nutritionism is a special case of misinterpretation and miscommunication of scientific results – something many data scientists encounter in their work.The wonderful world of recommender systemshttps://yanirseroussi.com/2015/10/02/the-wonderful-world-of-recommender-systems/Fri, 02 Oct 2015 05:25:57 +0000https://yanirseroussi.com/2015/10/02/the-wonderful-world-of-recommender-systems/Giving an overview of the field and common paradigms, and debunking five common myths about recommender systems.You don’t need a data scientist (yet)https://yanirseroussi.com/2015/08/24/you-dont-need-a-data-scientist-yet/Mon, 24 Aug 2015 08:25:30 +0000https://yanirseroussi.com/2015/08/24/you-dont-need-a-data-scientist-yet/Hiring data scientists prematurely is wasteful and frustrating. Here are some questions to ask before you hire your first data scientist.Learning about deep learning through album cover classificationhttps://yanirseroussi.com/2015/07/06/learning-about-deep-learning-through-album-cover-classification/Mon, 06 Jul 2015 22:21:42 +0000https://yanirseroussi.com/2015/07/06/learning-about-deep-learning-through-album-cover-classification/Progress on my album cover classification project, highlighting lessons that would be useful to others who are getting started with deep learning.Hopping on the deep learning bandwagonhttps://yanirseroussi.com/2015/06/06/hopping-on-the-deep-learning-bandwagon/Sat, 06 Jun 2015 05:00:22 +0000https://yanirseroussi.com/2015/06/06/hopping-on-the-deep-learning-bandwagon/To become proficient at solving data science problems, you need to get your hands dirty. Here, I used album cover classification to learn about deep learning.First steps in data science: author-aware sentiment analysishttps://yanirseroussi.com/2015/05/02/first-steps-in-data-science-author-aware-sentiment-analysis/Sat, 02 May 2015 08:31:10 +0000https://yanirseroussi.com/2015/05/02/first-steps-in-data-science-author-aware-sentiment-analysis/I became a data scientist by doing a PhD, but the same steps can be followed without a formal education program.My PhD workhttps://yanirseroussi.com/phd-work/Mon, 30 Mar 2015 03:23:33 +0000https://yanirseroussi.com/phd-work/An overview of my PhD in data science / artificial intelligence. Thesis title: Text Mining and Rating Prediction with Topical User Models.The long road to a lifestyle businesshttps://yanirseroussi.com/2015/03/22/the-long-road-to-a-lifestyle-business/Sun, 22 Mar 2015 09:43:47 +0000https://yanirseroussi.com/2015/03/22/the-long-road-to-a-lifestyle-business/Progress since leaving my last full-time job and setting on an independent path that includes data science consulting and work on my own projects.Learning to rank for personalised search (Yandex Search Personalisation – Kaggle Competition Summary – Part 2)https://yanirseroussi.com/2015/02/11/learning-to-rank-for-personalised-search-yandex-search-personalisation-kaggle-competition-summary-part-2/Wed, 11 Feb 2015 06:34:17 +0000https://yanirseroussi.com/2015/02/11/learning-to-rank-for-personalised-search-yandex-search-personalisation-kaggle-competition-summary-part-2/My team&rsquo;s solution to the Yandex Search Personalisation competition (finished 9th out of 194 teams).Is thinking like a search engine possible? (Yandex search personalisation – Kaggle competition summary – part 1)https://yanirseroussi.com/2015/01/29/is-thinking-like-a-search-engine-possible-yandex-search-personalisation-kaggle-competition-summary-part-1/Thu, 29 Jan 2015 10:37:39 +0000https://yanirseroussi.com/2015/01/29/is-thinking-like-a-search-engine-possible-yandex-search-personalisation-kaggle-competition-summary-part-1/Insights on search personalisation and SEO from participating in a Kaggle competition (finished 9th out of 194 teams).Stochastic Gradient Boosting: Choosing the Best Number of Iterationshttps://yanirseroussi.com/2014/12/29/stochastic-gradient-boosting-choosing-the-best-number-of-iterations/Mon, 29 Dec 2014 02:30:06 +0000https://yanirseroussi.com/2014/12/29/stochastic-gradient-boosting-choosing-the-best-number-of-iterations/Exploring an approach to choosing the optimal number of iterations in stochastic gradient boosting, following a bug I found in scikit-learn.Fitting noise: Forecasting the sale price of bulldozers (Kaggle competition summary)https://yanirseroussi.com/2014/11/19/fitting-noise-forecasting-the-sale-price-of-bulldozers-kaggle-competition-summary/Wed, 19 Nov 2014 09:17:34 +0000https://yanirseroussi.com/2014/11/19/fitting-noise-forecasting-the-sale-price-of-bulldozers-kaggle-competition-summary/Summary of a Kaggle competition to forecast bulldozer sale price, where I finished 9th out of 476 teams.What is data science?https://yanirseroussi.com/2014/10/23/what-is-data-science/Thu, 23 Oct 2014 03:22:08 +0000https://yanirseroussi.com/2014/10/23/what-is-data-science/Data science has been a hot term in the past few years. Still, there isn&rsquo;t a single definition of the field. This post discusses my favourite definition.Greek Media Monitoring Kaggle competition: My approachhttps://yanirseroussi.com/2014/10/07/greek-media-monitoring-kaggle-competition-my-approach/Tue, 07 Oct 2014 03:21:35 +0000https://yanirseroussi.com/2014/10/07/greek-media-monitoring-kaggle-competition-my-approach/Summary of my approach to the Greek Media Monitoring Kaggle competition, where I finished 6th out of 120 teams.Bandcamp recommendation and discovery algorithmshttps://yanirseroussi.com/2014/09/19/bandcamp-recommendation-and-discovery-algorithms/Fri, 19 Sep 2014 14:26:55 +0000https://yanirseroussi.com/2014/09/19/bandcamp-recommendation-and-discovery-algorithms/The recommendation backend for my BCRecommender service for personalised Bandcamp music discovery.How to (almost) win Kaggle competitionshttps://yanirseroussi.com/2014/08/24/how-to-almost-win-kaggle-competitions/Sun, 24 Aug 2014 12:40:53 +0000https://yanirseroussi.com/2014/08/24/how-to-almost-win-kaggle-competitions/Summary of a talk I gave at the Data Science Sydney meetup with ten tips on almost-winning Kaggle competitions.Data’s hierarchy of needshttps://yanirseroussi.com/2014/08/17/datas-hierarchy-of-needs/Sun, 17 Aug 2014 13:09:30 +0000https://yanirseroussi.com/2014/08/17/datas-hierarchy-of-needs/Discussing the hierarchy of needs proposed by Jay Kreps. Key takeaway: Data-driven algorithms &amp; insights can only be as good as the underlying data.Kaggle competition tips and summarieshttps://yanirseroussi.com/kaggle/Sat, 05 Apr 2014 23:46:10 +0000https://yanirseroussi.com/kaggle/Pointers to all my Kaggle advice posts and competition summaries.Kaggle beginner tipshttps://yanirseroussi.com/2014/01/19/kaggle-beginner-tips/Sun, 19 Jan 2014 10:34:28 +0000https://yanirseroussi.com/2014/01/19/kaggle-beginner-tips/First post! An email I sent to members of the Data Science Sydney Meetup with tips on how to get started with Kaggle competitions. \ No newline at end of file diff --git a/tags/data-strategy/index.html b/tags/data-strategy/index.html index e9c2f2ae9..9bb3330c8 100644 --- a/tags/data-strategy/index.html +++ b/tags/data-strategy/index.html @@ -2,7 +2,7 @@

      Data Strategy

      How to avoid startups with poor development processes

      Questions that prospective data specialists and engineers should ask about development processes before accepting a startup role.

      June 3, 2024

      Plumbing, Decisions, and Automation: De-hyping Data & AI

      Three essential questions to understand where an organisation stands when it comes to Data & AI (with zero hype).

      May 27, 2024

      Question startup culture before accepting a data-to-AI role

      Eight questions that prospective data-to-AI employees should ask about a startup’s work and data culture.

      May 20, 2024

      Probing the People aspects of an early-stage startup

      Ten questions that prospective employees should ask about a startup’s team, especially for data-centric roles.

      May 13, 2024

      Business questions to ask before taking a startup data role

      Fourteen questions that prospective employees should ask about a startup’s business model and product, especially for data-focused roles.

      May 6, 2024

      Assessing a startup's data-to-AI health

      Reviewing the areas that should be assessed to determine a startup’s opportunities and challenges on the data/AI/ML front.

      April 22, 2024

      My experience as a Data Tech Lead with Work on Climate

      The story of how I joined Work on Climate as a volunteer and became its data tech lead, with lessons applied to consulting & fractional work.

      April 8, 2024

      Two types of startup data problems

      Classifying startups as ML-centric or non-ML is a helpful exercise to uncover the data challenges they’re likely to face.

      March 4, 2024

      Avoiding AI complexity: First, write no code

      Two stories of getting AI functionality to production, which demonstrate the risks inherent in custom development versus starting with a no-code approach.

      February 26, 2024

      Building your startup's minimum viable data stack

      First post in a series on building a minimum viable data stack for startups, introducing key definitions, components, and considerations.

      February 19, 2024
      Text and figures licensed under CC BY-NC-ND 4.0 by Yanir Seroussi, except where noted otherwise  | +">

      Data Strategy

      Startup data health starts with healthy event tracking

      Expanding on the startup health check question of tracking Kukuyeva’s five business aspects as wide events.

      June 10, 2024

      How to avoid startups with poor development processes

      Questions that prospective data specialists and engineers should ask about development processes before accepting a startup role.

      June 3, 2024

      Plumbing, Decisions, and Automation: De-hyping Data & AI

      Three essential questions to understand where an organisation stands when it comes to Data & AI (with zero hype).

      May 27, 2024

      Question startup culture before accepting a data-to-AI role

      Eight questions that prospective data-to-AI employees should ask about a startup’s work and data culture.

      May 20, 2024

      Probing the People aspects of an early-stage startup

      Ten questions that prospective employees should ask about a startup’s team, especially for data-centric roles.

      May 13, 2024

      Business questions to ask before taking a startup data role

      Fourteen questions that prospective employees should ask about a startup’s business model and product, especially for data-focused roles.

      May 6, 2024

      Assessing a startup's data-to-AI health

      Reviewing the areas that should be assessed to determine a startup’s opportunities and challenges on the data/AI/ML front.

      April 22, 2024

      My experience as a Data Tech Lead with Work on Climate

      The story of how I joined Work on Climate as a volunteer and became its data tech lead, with lessons applied to consulting & fractional work.

      April 8, 2024

      Two types of startup data problems

      Classifying startups as ML-centric or non-ML is a helpful exercise to uncover the data challenges they’re likely to face.

      March 4, 2024

      Avoiding AI complexity: First, write no code

      Two stories of getting AI functionality to production, which demonstrate the risks inherent in custom development versus starting with a no-code approach.

      February 26, 2024

      Building your startup's minimum viable data stack

      First post in a series on building a minimum viable data stack for startups, introducing key definitions, components, and considerations.

      February 19, 2024
      Text and figures licensed under CC BY-NC-ND 4.0 by Yanir Seroussi, except where noted otherwise  | Powered by Hugo & PaperMod
      \ No newline at end of file diff --git a/tags/data-strategy/index.xml b/tags/data-strategy/index.xml index 1fa111e5a..bf5979340 100644 --- a/tags/data-strategy/index.xml +++ b/tags/data-strategy/index.xml @@ -1 +1 @@ -Data Strategy on Yanir Seroussi | Data & AI for Startup Impacthttps://yanirseroussi.com/tags/data-strategy/Recent content in Data Strategy on Yanir Seroussi | Data & AI for Startup ImpactHugo -- gohugo.ioen-auText and figures licensed under [CC BY-NC-ND 4.0](https://creativecommons.org/licenses/by-nc-nd/4.0/) by [Yanir Seroussi](https://yanirseroussi.com/about/), except where noted otherwiseMon, 03 Jun 2024 02:45:00 +0000How to avoid startups with poor development processeshttps://yanirseroussi.com/2024/06/03/how-to-avoid-startups-with-poor-development-processes/Mon, 03 Jun 2024 02:45:00 +0000https://yanirseroussi.com/2024/06/03/how-to-avoid-startups-with-poor-development-processes/Questions that prospective data specialists and engineers should ask about development processes before accepting a startup role.Plumbing, Decisions, and Automation: De-hyping Data & AIhttps://yanirseroussi.com/2024/05/27/plumbing-decisions-and-automation-de-hyping-data-and-ai/Mon, 27 May 2024 02:00:00 +0000https://yanirseroussi.com/2024/05/27/plumbing-decisions-and-automation-de-hyping-data-and-ai/Three essential questions to understand where an organisation stands when it comes to Data &amp; AI (with zero hype).Question startup culture before accepting a data-to-AI rolehttps://yanirseroussi.com/2024/05/20/question-startup-culture-before-accepting-a-data-to-ai-role/Mon, 20 May 2024 02:25:00 +0000https://yanirseroussi.com/2024/05/20/question-startup-culture-before-accepting-a-data-to-ai-role/Eight questions that prospective data-to-AI employees should ask about a startup&rsquo;s work and data culture.Probing the People aspects of an early-stage startuphttps://yanirseroussi.com/2024/05/13/probing-the-people-aspects-of-an-early-stage-startup/Mon, 13 May 2024 02:00:00 +0000https://yanirseroussi.com/2024/05/13/probing-the-people-aspects-of-an-early-stage-startup/Ten questions that prospective employees should ask about a startup&rsquo;s team, especially for data-centric roles.Business questions to ask before taking a startup data rolehttps://yanirseroussi.com/2024/05/06/business-questions-to-ask-before-taking-a-startup-data-role/Mon, 06 May 2024 04:30:00 +0000https://yanirseroussi.com/2024/05/06/business-questions-to-ask-before-taking-a-startup-data-role/Fourteen questions that prospective employees should ask about a startup&rsquo;s business model and product, especially for data-focused roles.Assessing a startup's data-to-AI healthhttps://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/Mon, 22 Apr 2024 06:00:00 +0000https://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/Reviewing the areas that should be assessed to determine a startup&rsquo;s opportunities and challenges on the data/AI/ML front.My experience as a Data Tech Lead with Work on Climatehttps://yanirseroussi.com/2024/04/08/my-experience-as-a-data-tech-lead-with-work-on-climate/Mon, 08 Apr 2024 02:00:00 +0000https://yanirseroussi.com/2024/04/08/my-experience-as-a-data-tech-lead-with-work-on-climate/The story of how I joined Work on Climate as a volunteer and became its data tech lead, with lessons applied to consulting &amp; fractional work.Two types of startup data problemshttps://yanirseroussi.com/2024/03/04/two-types-of-startup-data-problems/Mon, 04 Mar 2024 02:00:00 +0000https://yanirseroussi.com/2024/03/04/two-types-of-startup-data-problems/Classifying startups as ML-centric or non-ML is a helpful exercise to uncover the data challenges they&rsquo;re likely to face.Avoiding AI complexity: First, write no codehttps://yanirseroussi.com/2024/02/26/avoiding-ai-complexity-first-write-no-code/Mon, 26 Feb 2024 01:45:00 +0000https://yanirseroussi.com/2024/02/26/avoiding-ai-complexity-first-write-no-code/Two stories of getting AI functionality to production, which demonstrate the risks inherent in custom development versus starting with a no-code approach.Building your startup's minimum viable data stackhttps://yanirseroussi.com/2024/02/19/building-your-startups-minimum-viable-data-stack/Mon, 19 Feb 2024 00:00:00 +0000https://yanirseroussi.com/2024/02/19/building-your-startups-minimum-viable-data-stack/First post in a series on building a minimum viable data stack for startups, introducing key definitions, components, and considerations. \ No newline at end of file +Data Strategy on Yanir Seroussi | Data & AI for Startup Impacthttps://yanirseroussi.com/tags/data-strategy/Recent content in Data Strategy on Yanir Seroussi | Data & AI for Startup ImpactHugo -- gohugo.ioen-auText and figures licensed under [CC BY-NC-ND 4.0](https://creativecommons.org/licenses/by-nc-nd/4.0/) by [Yanir Seroussi](https://yanirseroussi.com/about/), except where noted otherwiseMon, 10 Jun 2024 04:00:00 +0000Startup data health starts with healthy event trackinghttps://yanirseroussi.com/2024/06/10/startup-data-health-starts-with-healthy-event-tracking/Mon, 10 Jun 2024 04:00:00 +0000https://yanirseroussi.com/2024/06/10/startup-data-health-starts-with-healthy-event-tracking/Expanding on the startup health check question of tracking Kukuyeva&rsquo;s five business aspects as wide events.How to avoid startups with poor development processeshttps://yanirseroussi.com/2024/06/03/how-to-avoid-startups-with-poor-development-processes/Mon, 03 Jun 2024 02:45:00 +0000https://yanirseroussi.com/2024/06/03/how-to-avoid-startups-with-poor-development-processes/Questions that prospective data specialists and engineers should ask about development processes before accepting a startup role.Plumbing, Decisions, and Automation: De-hyping Data & AIhttps://yanirseroussi.com/2024/05/27/plumbing-decisions-and-automation-de-hyping-data-and-ai/Mon, 27 May 2024 02:00:00 +0000https://yanirseroussi.com/2024/05/27/plumbing-decisions-and-automation-de-hyping-data-and-ai/Three essential questions to understand where an organisation stands when it comes to Data &amp; AI (with zero hype).Question startup culture before accepting a data-to-AI rolehttps://yanirseroussi.com/2024/05/20/question-startup-culture-before-accepting-a-data-to-ai-role/Mon, 20 May 2024 02:25:00 +0000https://yanirseroussi.com/2024/05/20/question-startup-culture-before-accepting-a-data-to-ai-role/Eight questions that prospective data-to-AI employees should ask about a startup&rsquo;s work and data culture.Probing the People aspects of an early-stage startuphttps://yanirseroussi.com/2024/05/13/probing-the-people-aspects-of-an-early-stage-startup/Mon, 13 May 2024 02:00:00 +0000https://yanirseroussi.com/2024/05/13/probing-the-people-aspects-of-an-early-stage-startup/Ten questions that prospective employees should ask about a startup&rsquo;s team, especially for data-centric roles.Business questions to ask before taking a startup data rolehttps://yanirseroussi.com/2024/05/06/business-questions-to-ask-before-taking-a-startup-data-role/Mon, 06 May 2024 04:30:00 +0000https://yanirseroussi.com/2024/05/06/business-questions-to-ask-before-taking-a-startup-data-role/Fourteen questions that prospective employees should ask about a startup&rsquo;s business model and product, especially for data-focused roles.Assessing a startup's data-to-AI healthhttps://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/Mon, 22 Apr 2024 06:00:00 +0000https://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/Reviewing the areas that should be assessed to determine a startup&rsquo;s opportunities and challenges on the data/AI/ML front.My experience as a Data Tech Lead with Work on Climatehttps://yanirseroussi.com/2024/04/08/my-experience-as-a-data-tech-lead-with-work-on-climate/Mon, 08 Apr 2024 02:00:00 +0000https://yanirseroussi.com/2024/04/08/my-experience-as-a-data-tech-lead-with-work-on-climate/The story of how I joined Work on Climate as a volunteer and became its data tech lead, with lessons applied to consulting &amp; fractional work.Two types of startup data problemshttps://yanirseroussi.com/2024/03/04/two-types-of-startup-data-problems/Mon, 04 Mar 2024 02:00:00 +0000https://yanirseroussi.com/2024/03/04/two-types-of-startup-data-problems/Classifying startups as ML-centric or non-ML is a helpful exercise to uncover the data challenges they&rsquo;re likely to face.Avoiding AI complexity: First, write no codehttps://yanirseroussi.com/2024/02/26/avoiding-ai-complexity-first-write-no-code/Mon, 26 Feb 2024 01:45:00 +0000https://yanirseroussi.com/2024/02/26/avoiding-ai-complexity-first-write-no-code/Two stories of getting AI functionality to production, which demonstrate the risks inherent in custom development versus starting with a no-code approach.Building your startup's minimum viable data stackhttps://yanirseroussi.com/2024/02/19/building-your-startups-minimum-viable-data-stack/Mon, 19 Feb 2024 00:00:00 +0000https://yanirseroussi.com/2024/02/19/building-your-startups-minimum-viable-data-stack/First post in a series on building a minimum viable data stack for startups, introducing key definitions, components, and considerations. \ No newline at end of file diff --git a/tags/index.html b/tags/index.html index 6585a528e..2ae82d8f4 100644 --- a/tags/index.html +++ b/tags/index.html @@ -2,7 +2,7 @@

      Tags

      Text and figures licensed under CC BY-NC-ND 4.0 by Yanir Seroussi, except where noted otherwise  | +">

      Tags

      Text and figures licensed under CC BY-NC-ND 4.0 by Yanir Seroussi, except where noted otherwise  | Powered by Hugo & PaperMod
      \ No newline at end of file diff --git a/tags/index.xml b/tags/index.xml index 93e3c76d7..b86dcc93a 100644 --- a/tags/index.xml +++ b/tags/index.xml @@ -1 +1 @@ -Tags on Yanir Seroussi | Data & AI for Startup Impacthttps://yanirseroussi.com/tags/Recent content in Tags on Yanir Seroussi | Data & AI for Startup ImpactHugo -- gohugo.ioen-auText and figures licensed under [CC BY-NC-ND 4.0](https://creativecommons.org/licenses/by-nc-nd/4.0/) by [Yanir Seroussi](https://yanirseroussi.com/about/), except where noted otherwiseMon, 03 Jun 2024 02:45:00 +0000Businesshttps://yanirseroussi.com/tags/business/Mon, 03 Jun 2024 02:45:00 +0000https://yanirseroussi.com/tags/business/Careerhttps://yanirseroussi.com/tags/career/Mon, 03 Jun 2024 02:45:00 +0000https://yanirseroussi.com/tags/career/Data Strategyhttps://yanirseroussi.com/tags/data-strategy/Mon, 03 Jun 2024 02:45:00 +0000https://yanirseroussi.com/tags/data-strategy/Software Engineeringhttps://yanirseroussi.com/tags/software-engineering/Mon, 03 Jun 2024 02:45:00 +0000https://yanirseroussi.com/tags/software-engineering/Startupshttps://yanirseroussi.com/tags/startups/Mon, 03 Jun 2024 02:45:00 +0000https://yanirseroussi.com/tags/startups/Artificial Intelligencehttps://yanirseroussi.com/tags/artificial-intelligence/Mon, 27 May 2024 02:00:00 +0000https://yanirseroussi.com/tags/artificial-intelligence/Data Engineeringhttps://yanirseroussi.com/tags/data-engineering/Mon, 27 May 2024 02:00:00 +0000https://yanirseroussi.com/tags/data-engineering/Data Sciencehttps://yanirseroussi.com/tags/data-science/Mon, 27 May 2024 02:00:00 +0000https://yanirseroussi.com/tags/data-science/Bookshttps://yanirseroussi.com/tags/books/Sat, 25 May 2024 00:00:00 +0000https://yanirseroussi.com/tags/books/Futurismhttps://yanirseroussi.com/tags/futurism/Sat, 25 May 2024 00:00:00 +0000https://yanirseroussi.com/tags/futurism/Quoteshttps://yanirseroussi.com/tags/quotes/Sat, 25 May 2024 00:00:00 +0000https://yanirseroussi.com/tags/quotes/Consultinghttps://yanirseroussi.com/tags/consulting/Mon, 29 Apr 2024 06:30:00 +0000https://yanirseroussi.com/tags/consulting/Personalhttps://yanirseroussi.com/tags/personal/Mon, 29 Apr 2024 06:30:00 +0000https://yanirseroussi.com/tags/personal/Analyticshttps://yanirseroussi.com/tags/analytics/Mon, 22 Apr 2024 06:00:00 +0000https://yanirseroussi.com/tags/analytics/Machine Learninghttps://yanirseroussi.com/tags/machine-learning/Mon, 22 Apr 2024 06:00:00 +0000https://yanirseroussi.com/tags/machine-learning/LinkedInhttps://yanirseroussi.com/tags/linkedin/Thu, 11 Apr 2024 01:45:25 +0000https://yanirseroussi.com/tags/linkedin/Marketinghttps://yanirseroussi.com/tags/marketing/Thu, 11 Apr 2024 01:45:25 +0000https://yanirseroussi.com/tags/marketing/Climate Changehttps://yanirseroussi.com/tags/climate-change/Mon, 08 Apr 2024 02:00:00 +0000https://yanirseroussi.com/tags/climate-change/Environmenthttps://yanirseroussi.com/tags/environment/Mon, 08 Apr 2024 02:00:00 +0000https://yanirseroussi.com/tags/environment/Remote Workhttps://yanirseroussi.com/tags/remote-work/Mon, 08 Apr 2024 02:00:00 +0000https://yanirseroussi.com/tags/remote-work/Marine Sciencehttps://yanirseroussi.com/tags/marine-science/Mon, 01 Apr 2024 06:00:00 +0000https://yanirseroussi.com/tags/marine-science/Reef Life Surveyhttps://yanirseroussi.com/tags/reef-life-survey/Mon, 01 Apr 2024 06:00:00 +0000https://yanirseroussi.com/tags/reef-life-survey/Productivityhttps://yanirseroussi.com/tags/productivity/Tue, 12 Mar 2024 06:19:31 +0000https://yanirseroussi.com/tags/productivity/Blogginghttps://yanirseroussi.com/tags/blogging/Fri, 19 Jan 2024 00:00:00 +0000https://yanirseroussi.com/tags/blogging/Data Businesshttps://yanirseroussi.com/tags/data-business/Mon, 18 Dec 2023 00:30:00 +0000https://yanirseroussi.com/tags/data-business/Energy Marketshttps://yanirseroussi.com/tags/energy-markets/Thu, 14 Dec 2023 00:15:00 +0000https://yanirseroussi.com/tags/energy-markets/Data Visualisationhttps://yanirseroussi.com/tags/data-visualisation/Wed, 29 Nov 2023 02:00:00 +0000https://yanirseroussi.com/tags/data-visualisation/Web Developmenthttps://yanirseroussi.com/tags/web-development/Wed, 29 Nov 2023 02:00:00 +0000https://yanirseroussi.com/tags/web-development/Ethicshttps://yanirseroussi.com/tags/ethics/Fri, 06 Oct 2023 05:00:00 +0000https://yanirseroussi.com/tags/ethics/DevOpshttps://yanirseroussi.com/tags/devops/Sun, 23 Jul 2023 00:07:15 +0000https://yanirseroussi.com/tags/devops/GitHubhttps://yanirseroussi.com/tags/github/Sun, 23 Jul 2023 00:07:15 +0000https://yanirseroussi.com/tags/github/Securityhttps://yanirseroussi.com/tags/security/Sun, 23 Jul 2023 00:07:15 +0000https://yanirseroussi.com/tags/security/Hugohttps://yanirseroussi.com/tags/hugo/Mon, 17 Jul 2023 00:06:15 +0000https://yanirseroussi.com/tags/hugo/Hackershttps://yanirseroussi.com/tags/hackers/Fri, 26 May 2023 00:03:00 +0000https://yanirseroussi.com/tags/hackers/Machine Intelligencehttps://yanirseroussi.com/tags/machine-intelligence/Fri, 21 Apr 2023 00:06:30 +0000https://yanirseroussi.com/tags/machine-intelligence/Causal Inferencehttps://yanirseroussi.com/tags/causal-inference/Mon, 12 Sep 2022 02:45:00 +0000https://yanirseroussi.com/tags/causal-inference/Automattichttps://yanirseroussi.com/tags/automattic/Mon, 06 Jun 2022 00:00:00 +0000https://yanirseroussi.com/tags/automattic/Orkestrahttps://yanirseroussi.com/tags/orkestra/Mon, 06 Jun 2022 00:00:00 +0000https://yanirseroussi.com/tags/orkestra/Politicshttps://yanirseroussi.com/tags/politics/Mon, 06 Jun 2022 00:00:00 +0000https://yanirseroussi.com/tags/politics/Sustainabilityhttps://yanirseroussi.com/tags/sustainability/Mon, 06 Jun 2022 00:00:00 +0000https://yanirseroussi.com/tags/sustainability/Deep Learninghttps://yanirseroussi.com/tags/deep-learning/Sun, 20 Mar 2022 04:30:00 +0000https://yanirseroussi.com/tags/deep-learning/Fast.aihttps://yanirseroussi.com/tags/fast.ai/Sun, 20 Mar 2022 04:30:00 +0000https://yanirseroussi.com/tags/fast.ai/Split Testinghttps://yanirseroussi.com/tags/split-testing/Fri, 14 Jan 2022 00:05:40 +0000https://yanirseroussi.com/tags/split-testing/Statisticshttps://yanirseroussi.com/tags/statistics/Fri, 14 Jan 2022 00:05:40 +0000https://yanirseroussi.com/tags/statistics/Cloudflarehttps://yanirseroussi.com/tags/cloudflare/Wed, 10 Nov 2021 06:30:00 +0000https://yanirseroussi.com/tags/cloudflare/WordPresshttps://yanirseroussi.com/tags/wordpress/Wed, 10 Nov 2021 06:30:00 +0000https://yanirseroussi.com/tags/wordpress/Bootstrappinghttps://yanirseroussi.com/tags/bootstrapping/Mon, 24 Aug 2020 01:35:17 +0000https://yanirseroussi.com/tags/bootstrapping/Confidence Intervalshttps://yanirseroussi.com/tags/confidence-intervals/Mon, 24 Aug 2020 01:35:17 +0000https://yanirseroussi.com/tags/confidence-intervals/Frequently Asked Questionshttps://yanirseroussi.com/tags/frequently-asked-questions/Sun, 15 Oct 2017 09:15:25 +0000https://yanirseroussi.com/tags/frequently-asked-questions/Bandcamphttps://yanirseroussi.com/tags/bandcamp/Sat, 02 Sep 2017 10:19:02 +0000https://yanirseroussi.com/tags/bandcamp/BCRecommenderhttps://yanirseroussi.com/tags/bcrecommender/Sat, 02 Sep 2017 10:19:02 +0000https://yanirseroussi.com/tags/bcrecommender/Elasticsearchhttps://yanirseroussi.com/tags/elasticsearch/Sat, 29 Jul 2017 05:39:26 +0000https://yanirseroussi.com/tags/elasticsearch/JavaScripthttps://yanirseroussi.com/tags/javascript/Sat, 03 Jun 2017 00:49:05 +0000https://yanirseroussi.com/tags/javascript/Predictive Modellinghttps://yanirseroussi.com/tags/predictive-modelling/Sun, 08 Jan 2017 20:02:30 +0000https://yanirseroussi.com/tags/predictive-modelling/Science Communicationhttps://yanirseroussi.com/tags/science-communication/Sun, 08 Jan 2017 20:02:30 +0000https://yanirseroussi.com/tags/science-communication/Search Engine Optimisationhttps://yanirseroussi.com/tags/search-engine-optimisation/Sun, 08 Jan 2017 20:02:30 +0000https://yanirseroussi.com/tags/search-engine-optimisation/Insightshttps://yanirseroussi.com/tags/insights/Mon, 19 Sep 2016 21:28:44 +0000https://yanirseroussi.com/tags/insights/Economicshttps://yanirseroussi.com/tags/economics/Sun, 20 Mar 2016 20:33:43 +0000https://yanirseroussi.com/tags/economics/Scuba Divinghttps://yanirseroussi.com/tags/scuba-diving/Sun, 24 Jan 2016 00:32:25 +0000https://yanirseroussi.com/tags/scuba-diving/Facebookhttps://yanirseroussi.com/tags/facebook/Tue, 08 Dec 2015 06:57:25 +0000https://yanirseroussi.com/tags/facebook/Kagglehttps://yanirseroussi.com/tags/kaggle/Mon, 23 Nov 2015 04:14:21 +0000https://yanirseroussi.com/tags/kaggle/MongoDBhttps://yanirseroussi.com/tags/mongodb/Wed, 04 Nov 2015 03:53:18 +0000https://yanirseroussi.com/tags/mongodb/Healthhttps://yanirseroussi.com/tags/health/Mon, 19 Oct 2015 00:02:32 +0000https://yanirseroussi.com/tags/health/Nutritionhttps://yanirseroussi.com/tags/nutrition/Mon, 19 Oct 2015 00:02:32 +0000https://yanirseroussi.com/tags/nutrition/Nutritionismhttps://yanirseroussi.com/tags/nutritionism/Mon, 19 Oct 2015 00:02:32 +0000https://yanirseroussi.com/tags/nutritionism/Recommender Systemshttps://yanirseroussi.com/tags/recommender-systems/Fri, 02 Oct 2015 05:25:57 +0000https://yanirseroussi.com/tags/recommender-systems/Parse.comhttps://yanirseroussi.com/tags/parse.com/Fri, 31 Jul 2015 03:29:50 +0000https://yanirseroussi.com/tags/parse.com/Sentiment Analysishttps://yanirseroussi.com/tags/sentiment-analysis/Sat, 02 May 2015 08:31:10 +0000https://yanirseroussi.com/tags/sentiment-analysis/Divestmenthttps://yanirseroussi.com/tags/divestment/Fri, 24 Apr 2015 00:19:36 +0000https://yanirseroussi.com/tags/divestment/Fossil Fuelshttps://yanirseroussi.com/tags/fossil-fuels/Fri, 24 Apr 2015 00:19:36 +0000https://yanirseroussi.com/tags/fossil-fuels/Gradient Boostinghttps://yanirseroussi.com/tags/gradient-boosting/Wed, 11 Feb 2015 06:34:17 +0000https://yanirseroussi.com/tags/gradient-boosting/Kaggle Competitionhttps://yanirseroussi.com/tags/kaggle-competition/Wed, 11 Feb 2015 06:34:17 +0000https://yanirseroussi.com/tags/kaggle-competition/PhantomJShttps://yanirseroussi.com/tags/phantomjs/Thu, 15 Jan 2015 04:41:16 +0000https://yanirseroussi.com/tags/phantomjs/Scikit-Learnhttps://yanirseroussi.com/tags/scikit-learn/Mon, 29 Dec 2014 02:30:06 +0000https://yanirseroussi.com/tags/scikit-learn/Traction Bookhttps://yanirseroussi.com/tags/traction-book/Mon, 15 Dec 2014 04:25:25 +0000https://yanirseroussi.com/tags/traction-book/Price Forecastinghttps://yanirseroussi.com/tags/price-forecasting/Wed, 19 Nov 2014 09:17:34 +0000https://yanirseroussi.com/tags/price-forecasting/Musichttps://yanirseroussi.com/tags/music/Wed, 05 Nov 2014 02:29:35 +0000https://yanirseroussi.com/tags/music/Multi-Label Classificationhttps://yanirseroussi.com/tags/multi-label-classification/Tue, 07 Oct 2014 03:21:35 +0000https://yanirseroussi.com/tags/multi-label-classification/Music Industryhttps://yanirseroussi.com/tags/music-industry/Sat, 30 Aug 2014 08:11:38 +0000https://yanirseroussi.com/tags/music-industry/Kaggle Beginnershttps://yanirseroussi.com/tags/kaggle-beginners/Sun, 24 Aug 2014 12:40:53 +0000https://yanirseroussi.com/tags/kaggle-beginners/ \ No newline at end of file +Tags on Yanir Seroussi | Data & AI for Startup Impacthttps://yanirseroussi.com/tags/Recent content in Tags on Yanir Seroussi | Data & AI for Startup ImpactHugo -- gohugo.ioen-auText and figures licensed under [CC BY-NC-ND 4.0](https://creativecommons.org/licenses/by-nc-nd/4.0/) by [Yanir Seroussi](https://yanirseroussi.com/about/), except where noted otherwiseMon, 10 Jun 2024 04:00:00 +0000Analyticshttps://yanirseroussi.com/tags/analytics/Mon, 10 Jun 2024 04:00:00 +0000https://yanirseroussi.com/tags/analytics/Businesshttps://yanirseroussi.com/tags/business/Mon, 10 Jun 2024 04:00:00 +0000https://yanirseroussi.com/tags/business/Data Sciencehttps://yanirseroussi.com/tags/data-science/Mon, 10 Jun 2024 04:00:00 +0000https://yanirseroussi.com/tags/data-science/Data Strategyhttps://yanirseroussi.com/tags/data-strategy/Mon, 10 Jun 2024 04:00:00 +0000https://yanirseroussi.com/tags/data-strategy/Startupshttps://yanirseroussi.com/tags/startups/Mon, 10 Jun 2024 04:00:00 +0000https://yanirseroussi.com/tags/startups/Careerhttps://yanirseroussi.com/tags/career/Mon, 03 Jun 2024 02:45:00 +0000https://yanirseroussi.com/tags/career/Software Engineeringhttps://yanirseroussi.com/tags/software-engineering/Mon, 03 Jun 2024 02:45:00 +0000https://yanirseroussi.com/tags/software-engineering/Artificial Intelligencehttps://yanirseroussi.com/tags/artificial-intelligence/Mon, 27 May 2024 02:00:00 +0000https://yanirseroussi.com/tags/artificial-intelligence/Data Engineeringhttps://yanirseroussi.com/tags/data-engineering/Mon, 27 May 2024 02:00:00 +0000https://yanirseroussi.com/tags/data-engineering/Bookshttps://yanirseroussi.com/tags/books/Sat, 25 May 2024 00:00:00 +0000https://yanirseroussi.com/tags/books/Futurismhttps://yanirseroussi.com/tags/futurism/Sat, 25 May 2024 00:00:00 +0000https://yanirseroussi.com/tags/futurism/Quoteshttps://yanirseroussi.com/tags/quotes/Sat, 25 May 2024 00:00:00 +0000https://yanirseroussi.com/tags/quotes/Consultinghttps://yanirseroussi.com/tags/consulting/Mon, 29 Apr 2024 06:30:00 +0000https://yanirseroussi.com/tags/consulting/Personalhttps://yanirseroussi.com/tags/personal/Mon, 29 Apr 2024 06:30:00 +0000https://yanirseroussi.com/tags/personal/Machine Learninghttps://yanirseroussi.com/tags/machine-learning/Mon, 22 Apr 2024 06:00:00 +0000https://yanirseroussi.com/tags/machine-learning/LinkedInhttps://yanirseroussi.com/tags/linkedin/Thu, 11 Apr 2024 01:45:25 +0000https://yanirseroussi.com/tags/linkedin/Marketinghttps://yanirseroussi.com/tags/marketing/Thu, 11 Apr 2024 01:45:25 +0000https://yanirseroussi.com/tags/marketing/Climate Changehttps://yanirseroussi.com/tags/climate-change/Mon, 08 Apr 2024 02:00:00 +0000https://yanirseroussi.com/tags/climate-change/Environmenthttps://yanirseroussi.com/tags/environment/Mon, 08 Apr 2024 02:00:00 +0000https://yanirseroussi.com/tags/environment/Remote Workhttps://yanirseroussi.com/tags/remote-work/Mon, 08 Apr 2024 02:00:00 +0000https://yanirseroussi.com/tags/remote-work/Marine Sciencehttps://yanirseroussi.com/tags/marine-science/Mon, 01 Apr 2024 06:00:00 +0000https://yanirseroussi.com/tags/marine-science/Reef Life Surveyhttps://yanirseroussi.com/tags/reef-life-survey/Mon, 01 Apr 2024 06:00:00 +0000https://yanirseroussi.com/tags/reef-life-survey/Productivityhttps://yanirseroussi.com/tags/productivity/Tue, 12 Mar 2024 06:19:31 +0000https://yanirseroussi.com/tags/productivity/Blogginghttps://yanirseroussi.com/tags/blogging/Fri, 19 Jan 2024 00:00:00 +0000https://yanirseroussi.com/tags/blogging/Data Businesshttps://yanirseroussi.com/tags/data-business/Mon, 18 Dec 2023 00:30:00 +0000https://yanirseroussi.com/tags/data-business/Energy Marketshttps://yanirseroussi.com/tags/energy-markets/Thu, 14 Dec 2023 00:15:00 +0000https://yanirseroussi.com/tags/energy-markets/Data Visualisationhttps://yanirseroussi.com/tags/data-visualisation/Wed, 29 Nov 2023 02:00:00 +0000https://yanirseroussi.com/tags/data-visualisation/Web Developmenthttps://yanirseroussi.com/tags/web-development/Wed, 29 Nov 2023 02:00:00 +0000https://yanirseroussi.com/tags/web-development/Ethicshttps://yanirseroussi.com/tags/ethics/Fri, 06 Oct 2023 05:00:00 +0000https://yanirseroussi.com/tags/ethics/DevOpshttps://yanirseroussi.com/tags/devops/Sun, 23 Jul 2023 00:07:15 +0000https://yanirseroussi.com/tags/devops/GitHubhttps://yanirseroussi.com/tags/github/Sun, 23 Jul 2023 00:07:15 +0000https://yanirseroussi.com/tags/github/Securityhttps://yanirseroussi.com/tags/security/Sun, 23 Jul 2023 00:07:15 +0000https://yanirseroussi.com/tags/security/Hugohttps://yanirseroussi.com/tags/hugo/Mon, 17 Jul 2023 00:06:15 +0000https://yanirseroussi.com/tags/hugo/Hackershttps://yanirseroussi.com/tags/hackers/Fri, 26 May 2023 00:03:00 +0000https://yanirseroussi.com/tags/hackers/Machine Intelligencehttps://yanirseroussi.com/tags/machine-intelligence/Fri, 21 Apr 2023 00:06:30 +0000https://yanirseroussi.com/tags/machine-intelligence/Causal Inferencehttps://yanirseroussi.com/tags/causal-inference/Mon, 12 Sep 2022 02:45:00 +0000https://yanirseroussi.com/tags/causal-inference/Automattichttps://yanirseroussi.com/tags/automattic/Mon, 06 Jun 2022 00:00:00 +0000https://yanirseroussi.com/tags/automattic/Orkestrahttps://yanirseroussi.com/tags/orkestra/Mon, 06 Jun 2022 00:00:00 +0000https://yanirseroussi.com/tags/orkestra/Politicshttps://yanirseroussi.com/tags/politics/Mon, 06 Jun 2022 00:00:00 +0000https://yanirseroussi.com/tags/politics/Sustainabilityhttps://yanirseroussi.com/tags/sustainability/Mon, 06 Jun 2022 00:00:00 +0000https://yanirseroussi.com/tags/sustainability/Deep Learninghttps://yanirseroussi.com/tags/deep-learning/Sun, 20 Mar 2022 04:30:00 +0000https://yanirseroussi.com/tags/deep-learning/Fast.aihttps://yanirseroussi.com/tags/fast.ai/Sun, 20 Mar 2022 04:30:00 +0000https://yanirseroussi.com/tags/fast.ai/Split Testinghttps://yanirseroussi.com/tags/split-testing/Fri, 14 Jan 2022 00:05:40 +0000https://yanirseroussi.com/tags/split-testing/Statisticshttps://yanirseroussi.com/tags/statistics/Fri, 14 Jan 2022 00:05:40 +0000https://yanirseroussi.com/tags/statistics/Cloudflarehttps://yanirseroussi.com/tags/cloudflare/Wed, 10 Nov 2021 06:30:00 +0000https://yanirseroussi.com/tags/cloudflare/WordPresshttps://yanirseroussi.com/tags/wordpress/Wed, 10 Nov 2021 06:30:00 +0000https://yanirseroussi.com/tags/wordpress/Bootstrappinghttps://yanirseroussi.com/tags/bootstrapping/Mon, 24 Aug 2020 01:35:17 +0000https://yanirseroussi.com/tags/bootstrapping/Confidence Intervalshttps://yanirseroussi.com/tags/confidence-intervals/Mon, 24 Aug 2020 01:35:17 +0000https://yanirseroussi.com/tags/confidence-intervals/Frequently Asked Questionshttps://yanirseroussi.com/tags/frequently-asked-questions/Sun, 15 Oct 2017 09:15:25 +0000https://yanirseroussi.com/tags/frequently-asked-questions/Bandcamphttps://yanirseroussi.com/tags/bandcamp/Sat, 02 Sep 2017 10:19:02 +0000https://yanirseroussi.com/tags/bandcamp/BCRecommenderhttps://yanirseroussi.com/tags/bcrecommender/Sat, 02 Sep 2017 10:19:02 +0000https://yanirseroussi.com/tags/bcrecommender/Elasticsearchhttps://yanirseroussi.com/tags/elasticsearch/Sat, 29 Jul 2017 05:39:26 +0000https://yanirseroussi.com/tags/elasticsearch/JavaScripthttps://yanirseroussi.com/tags/javascript/Sat, 03 Jun 2017 00:49:05 +0000https://yanirseroussi.com/tags/javascript/Predictive Modellinghttps://yanirseroussi.com/tags/predictive-modelling/Sun, 08 Jan 2017 20:02:30 +0000https://yanirseroussi.com/tags/predictive-modelling/Science Communicationhttps://yanirseroussi.com/tags/science-communication/Sun, 08 Jan 2017 20:02:30 +0000https://yanirseroussi.com/tags/science-communication/Search Engine Optimisationhttps://yanirseroussi.com/tags/search-engine-optimisation/Sun, 08 Jan 2017 20:02:30 +0000https://yanirseroussi.com/tags/search-engine-optimisation/Insightshttps://yanirseroussi.com/tags/insights/Mon, 19 Sep 2016 21:28:44 +0000https://yanirseroussi.com/tags/insights/Economicshttps://yanirseroussi.com/tags/economics/Sun, 20 Mar 2016 20:33:43 +0000https://yanirseroussi.com/tags/economics/Scuba Divinghttps://yanirseroussi.com/tags/scuba-diving/Sun, 24 Jan 2016 00:32:25 +0000https://yanirseroussi.com/tags/scuba-diving/Facebookhttps://yanirseroussi.com/tags/facebook/Tue, 08 Dec 2015 06:57:25 +0000https://yanirseroussi.com/tags/facebook/Kagglehttps://yanirseroussi.com/tags/kaggle/Mon, 23 Nov 2015 04:14:21 +0000https://yanirseroussi.com/tags/kaggle/MongoDBhttps://yanirseroussi.com/tags/mongodb/Wed, 04 Nov 2015 03:53:18 +0000https://yanirseroussi.com/tags/mongodb/Healthhttps://yanirseroussi.com/tags/health/Mon, 19 Oct 2015 00:02:32 +0000https://yanirseroussi.com/tags/health/Nutritionhttps://yanirseroussi.com/tags/nutrition/Mon, 19 Oct 2015 00:02:32 +0000https://yanirseroussi.com/tags/nutrition/Nutritionismhttps://yanirseroussi.com/tags/nutritionism/Mon, 19 Oct 2015 00:02:32 +0000https://yanirseroussi.com/tags/nutritionism/Recommender Systemshttps://yanirseroussi.com/tags/recommender-systems/Fri, 02 Oct 2015 05:25:57 +0000https://yanirseroussi.com/tags/recommender-systems/Parse.comhttps://yanirseroussi.com/tags/parse.com/Fri, 31 Jul 2015 03:29:50 +0000https://yanirseroussi.com/tags/parse.com/Sentiment Analysishttps://yanirseroussi.com/tags/sentiment-analysis/Sat, 02 May 2015 08:31:10 +0000https://yanirseroussi.com/tags/sentiment-analysis/Divestmenthttps://yanirseroussi.com/tags/divestment/Fri, 24 Apr 2015 00:19:36 +0000https://yanirseroussi.com/tags/divestment/Fossil Fuelshttps://yanirseroussi.com/tags/fossil-fuels/Fri, 24 Apr 2015 00:19:36 +0000https://yanirseroussi.com/tags/fossil-fuels/Gradient Boostinghttps://yanirseroussi.com/tags/gradient-boosting/Wed, 11 Feb 2015 06:34:17 +0000https://yanirseroussi.com/tags/gradient-boosting/Kaggle Competitionhttps://yanirseroussi.com/tags/kaggle-competition/Wed, 11 Feb 2015 06:34:17 +0000https://yanirseroussi.com/tags/kaggle-competition/PhantomJShttps://yanirseroussi.com/tags/phantomjs/Thu, 15 Jan 2015 04:41:16 +0000https://yanirseroussi.com/tags/phantomjs/Scikit-Learnhttps://yanirseroussi.com/tags/scikit-learn/Mon, 29 Dec 2014 02:30:06 +0000https://yanirseroussi.com/tags/scikit-learn/Traction Bookhttps://yanirseroussi.com/tags/traction-book/Mon, 15 Dec 2014 04:25:25 +0000https://yanirseroussi.com/tags/traction-book/Price Forecastinghttps://yanirseroussi.com/tags/price-forecasting/Wed, 19 Nov 2014 09:17:34 +0000https://yanirseroussi.com/tags/price-forecasting/Musichttps://yanirseroussi.com/tags/music/Wed, 05 Nov 2014 02:29:35 +0000https://yanirseroussi.com/tags/music/Multi-Label Classificationhttps://yanirseroussi.com/tags/multi-label-classification/Tue, 07 Oct 2014 03:21:35 +0000https://yanirseroussi.com/tags/multi-label-classification/Music Industryhttps://yanirseroussi.com/tags/music-industry/Sat, 30 Aug 2014 08:11:38 +0000https://yanirseroussi.com/tags/music-industry/Kaggle Beginnershttps://yanirseroussi.com/tags/kaggle-beginners/Sun, 24 Aug 2014 12:40:53 +0000https://yanirseroussi.com/tags/kaggle-beginners/ \ No newline at end of file diff --git a/tags/startups/index.html b/tags/startups/index.html index f846a8784..cf83980d2 100644 --- a/tags/startups/index.html +++ b/tags/startups/index.html @@ -2,7 +2,7 @@

      Startups

      How to avoid startups with poor development processes

      Questions that prospective data specialists and engineers should ask about development processes before accepting a startup role.

      June 3, 2024

      Plumbing, Decisions, and Automation: De-hyping Data & AI

      Three essential questions to understand where an organisation stands when it comes to Data & AI (with zero hype).

      May 27, 2024

      Question startup culture before accepting a data-to-AI role

      Eight questions that prospective data-to-AI employees should ask about a startup’s work and data culture.

      May 20, 2024

      Probing the People aspects of an early-stage startup

      Ten questions that prospective employees should ask about a startup’s team, especially for data-centric roles.

      May 13, 2024

      Business questions to ask before taking a startup data role

      Fourteen questions that prospective employees should ask about a startup’s business model and product, especially for data-focused roles.

      May 6, 2024

      Mentorship and the art of actionable advice

      Reflections on what it takes to package expertise and deliver timely, actionable advice outside the context of employee relationships.

      April 29, 2024

      Assessing a startup's data-to-AI health

      Reviewing the areas that should be assessed to determine a startup’s opportunities and challenges on the data/AI/ML front.

      April 22, 2024

      My experience as a Data Tech Lead with Work on Climate

      The story of how I joined Work on Climate as a volunteer and became its data tech lead, with lessons applied to consulting & fractional work.

      April 8, 2024

      Two types of startup data problems

      Classifying startups as ML-centric or non-ML is a helpful exercise to uncover the data challenges they’re likely to face.

      March 4, 2024

      Avoiding AI complexity: First, write no code

      Two stories of getting AI functionality to production, which demonstrate the risks inherent in custom development versus starting with a no-code approach.

      February 26, 2024

      Building your startup's minimum viable data stack

      First post in a series on building a minimum viable data stack for startups, introducing key definitions, components, and considerations.

      February 19, 2024

      Substance over titles: Your first data hire may be a data scientist

      Advice for hiring a startup’s first data person: match skills to business needs, consider contractors, and get help from data people.

      February 5, 2024
      Text and figures licensed under CC BY-NC-ND 4.0 by Yanir Seroussi, except where noted otherwise  | +">

      Startups

      Startup data health starts with healthy event tracking

      Expanding on the startup health check question of tracking Kukuyeva’s five business aspects as wide events.

      June 10, 2024

      How to avoid startups with poor development processes

      Questions that prospective data specialists and engineers should ask about development processes before accepting a startup role.

      June 3, 2024

      Plumbing, Decisions, and Automation: De-hyping Data & AI

      Three essential questions to understand where an organisation stands when it comes to Data & AI (with zero hype).

      May 27, 2024

      Question startup culture before accepting a data-to-AI role

      Eight questions that prospective data-to-AI employees should ask about a startup’s work and data culture.

      May 20, 2024

      Probing the People aspects of an early-stage startup

      Ten questions that prospective employees should ask about a startup’s team, especially for data-centric roles.

      May 13, 2024

      Business questions to ask before taking a startup data role

      Fourteen questions that prospective employees should ask about a startup’s business model and product, especially for data-focused roles.

      May 6, 2024

      Mentorship and the art of actionable advice

      Reflections on what it takes to package expertise and deliver timely, actionable advice outside the context of employee relationships.

      April 29, 2024

      Assessing a startup's data-to-AI health

      Reviewing the areas that should be assessed to determine a startup’s opportunities and challenges on the data/AI/ML front.

      April 22, 2024

      My experience as a Data Tech Lead with Work on Climate

      The story of how I joined Work on Climate as a volunteer and became its data tech lead, with lessons applied to consulting & fractional work.

      April 8, 2024

      Two types of startup data problems

      Classifying startups as ML-centric or non-ML is a helpful exercise to uncover the data challenges they’re likely to face.

      March 4, 2024

      Avoiding AI complexity: First, write no code

      Two stories of getting AI functionality to production, which demonstrate the risks inherent in custom development versus starting with a no-code approach.

      February 26, 2024

      Building your startup's minimum viable data stack

      First post in a series on building a minimum viable data stack for startups, introducing key definitions, components, and considerations.

      February 19, 2024

      Substance over titles: Your first data hire may be a data scientist

      Advice for hiring a startup’s first data person: match skills to business needs, consider contractors, and get help from data people.

      February 5, 2024
      Text and figures licensed under CC BY-NC-ND 4.0 by Yanir Seroussi, except where noted otherwise  | Powered by Hugo & PaperMod
      \ No newline at end of file diff --git a/tags/startups/index.xml b/tags/startups/index.xml index a626b8880..10edae7cf 100644 --- a/tags/startups/index.xml +++ b/tags/startups/index.xml @@ -1 +1 @@ -Startups on Yanir Seroussi | Data & AI for Startup Impacthttps://yanirseroussi.com/tags/startups/Recent content in Startups on Yanir Seroussi | Data & AI for Startup ImpactHugo -- gohugo.ioen-auText and figures licensed under [CC BY-NC-ND 4.0](https://creativecommons.org/licenses/by-nc-nd/4.0/) by [Yanir Seroussi](https://yanirseroussi.com/about/), except where noted otherwiseMon, 03 Jun 2024 02:45:00 +0000How to avoid startups with poor development processeshttps://yanirseroussi.com/2024/06/03/how-to-avoid-startups-with-poor-development-processes/Mon, 03 Jun 2024 02:45:00 +0000https://yanirseroussi.com/2024/06/03/how-to-avoid-startups-with-poor-development-processes/Questions that prospective data specialists and engineers should ask about development processes before accepting a startup role.Plumbing, Decisions, and Automation: De-hyping Data & AIhttps://yanirseroussi.com/2024/05/27/plumbing-decisions-and-automation-de-hyping-data-and-ai/Mon, 27 May 2024 02:00:00 +0000https://yanirseroussi.com/2024/05/27/plumbing-decisions-and-automation-de-hyping-data-and-ai/Three essential questions to understand where an organisation stands when it comes to Data &amp; AI (with zero hype).Question startup culture before accepting a data-to-AI rolehttps://yanirseroussi.com/2024/05/20/question-startup-culture-before-accepting-a-data-to-ai-role/Mon, 20 May 2024 02:25:00 +0000https://yanirseroussi.com/2024/05/20/question-startup-culture-before-accepting-a-data-to-ai-role/Eight questions that prospective data-to-AI employees should ask about a startup&rsquo;s work and data culture.Probing the People aspects of an early-stage startuphttps://yanirseroussi.com/2024/05/13/probing-the-people-aspects-of-an-early-stage-startup/Mon, 13 May 2024 02:00:00 +0000https://yanirseroussi.com/2024/05/13/probing-the-people-aspects-of-an-early-stage-startup/Ten questions that prospective employees should ask about a startup&rsquo;s team, especially for data-centric roles.Business questions to ask before taking a startup data rolehttps://yanirseroussi.com/2024/05/06/business-questions-to-ask-before-taking-a-startup-data-role/Mon, 06 May 2024 04:30:00 +0000https://yanirseroussi.com/2024/05/06/business-questions-to-ask-before-taking-a-startup-data-role/Fourteen questions that prospective employees should ask about a startup&rsquo;s business model and product, especially for data-focused roles.Mentorship and the art of actionable advicehttps://yanirseroussi.com/2024/04/29/mentorship-and-the-art-of-actionable-advice/Mon, 29 Apr 2024 06:30:00 +0000https://yanirseroussi.com/2024/04/29/mentorship-and-the-art-of-actionable-advice/Reflections on what it takes to package expertise and deliver timely, actionable advice outside the context of employee relationships.Assessing a startup's data-to-AI healthhttps://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/Mon, 22 Apr 2024 06:00:00 +0000https://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/Reviewing the areas that should be assessed to determine a startup&rsquo;s opportunities and challenges on the data/AI/ML front.My experience as a Data Tech Lead with Work on Climatehttps://yanirseroussi.com/2024/04/08/my-experience-as-a-data-tech-lead-with-work-on-climate/Mon, 08 Apr 2024 02:00:00 +0000https://yanirseroussi.com/2024/04/08/my-experience-as-a-data-tech-lead-with-work-on-climate/The story of how I joined Work on Climate as a volunteer and became its data tech lead, with lessons applied to consulting &amp; fractional work.Two types of startup data problemshttps://yanirseroussi.com/2024/03/04/two-types-of-startup-data-problems/Mon, 04 Mar 2024 02:00:00 +0000https://yanirseroussi.com/2024/03/04/two-types-of-startup-data-problems/Classifying startups as ML-centric or non-ML is a helpful exercise to uncover the data challenges they&rsquo;re likely to face.Avoiding AI complexity: First, write no codehttps://yanirseroussi.com/2024/02/26/avoiding-ai-complexity-first-write-no-code/Mon, 26 Feb 2024 01:45:00 +0000https://yanirseroussi.com/2024/02/26/avoiding-ai-complexity-first-write-no-code/Two stories of getting AI functionality to production, which demonstrate the risks inherent in custom development versus starting with a no-code approach.Building your startup's minimum viable data stackhttps://yanirseroussi.com/2024/02/19/building-your-startups-minimum-viable-data-stack/Mon, 19 Feb 2024 00:00:00 +0000https://yanirseroussi.com/2024/02/19/building-your-startups-minimum-viable-data-stack/First post in a series on building a minimum viable data stack for startups, introducing key definitions, components, and considerations.Substance over titles: Your first data hire may be a data scientisthttps://yanirseroussi.com/2024/02/05/substance-over-titles-your-first-data-hire-may-be-a-data-scientist/Mon, 05 Feb 2024 02:45:00 +0000https://yanirseroussi.com/2024/02/05/substance-over-titles-your-first-data-hire-may-be-a-data-scientist/Advice for hiring a startup&rsquo;s first data person: match skills to business needs, consider contractors, and get help from data people. \ No newline at end of file +Startups on Yanir Seroussi | Data & AI for Startup Impacthttps://yanirseroussi.com/tags/startups/Recent content in Startups on Yanir Seroussi | Data & AI for Startup ImpactHugo -- gohugo.ioen-auText and figures licensed under [CC BY-NC-ND 4.0](https://creativecommons.org/licenses/by-nc-nd/4.0/) by [Yanir Seroussi](https://yanirseroussi.com/about/), except where noted otherwiseMon, 10 Jun 2024 04:00:00 +0000Startup data health starts with healthy event trackinghttps://yanirseroussi.com/2024/06/10/startup-data-health-starts-with-healthy-event-tracking/Mon, 10 Jun 2024 04:00:00 +0000https://yanirseroussi.com/2024/06/10/startup-data-health-starts-with-healthy-event-tracking/Expanding on the startup health check question of tracking Kukuyeva&rsquo;s five business aspects as wide events.How to avoid startups with poor development processeshttps://yanirseroussi.com/2024/06/03/how-to-avoid-startups-with-poor-development-processes/Mon, 03 Jun 2024 02:45:00 +0000https://yanirseroussi.com/2024/06/03/how-to-avoid-startups-with-poor-development-processes/Questions that prospective data specialists and engineers should ask about development processes before accepting a startup role.Plumbing, Decisions, and Automation: De-hyping Data & AIhttps://yanirseroussi.com/2024/05/27/plumbing-decisions-and-automation-de-hyping-data-and-ai/Mon, 27 May 2024 02:00:00 +0000https://yanirseroussi.com/2024/05/27/plumbing-decisions-and-automation-de-hyping-data-and-ai/Three essential questions to understand where an organisation stands when it comes to Data &amp; AI (with zero hype).Question startup culture before accepting a data-to-AI rolehttps://yanirseroussi.com/2024/05/20/question-startup-culture-before-accepting-a-data-to-ai-role/Mon, 20 May 2024 02:25:00 +0000https://yanirseroussi.com/2024/05/20/question-startup-culture-before-accepting-a-data-to-ai-role/Eight questions that prospective data-to-AI employees should ask about a startup&rsquo;s work and data culture.Probing the People aspects of an early-stage startuphttps://yanirseroussi.com/2024/05/13/probing-the-people-aspects-of-an-early-stage-startup/Mon, 13 May 2024 02:00:00 +0000https://yanirseroussi.com/2024/05/13/probing-the-people-aspects-of-an-early-stage-startup/Ten questions that prospective employees should ask about a startup&rsquo;s team, especially for data-centric roles.Business questions to ask before taking a startup data rolehttps://yanirseroussi.com/2024/05/06/business-questions-to-ask-before-taking-a-startup-data-role/Mon, 06 May 2024 04:30:00 +0000https://yanirseroussi.com/2024/05/06/business-questions-to-ask-before-taking-a-startup-data-role/Fourteen questions that prospective employees should ask about a startup&rsquo;s business model and product, especially for data-focused roles.Mentorship and the art of actionable advicehttps://yanirseroussi.com/2024/04/29/mentorship-and-the-art-of-actionable-advice/Mon, 29 Apr 2024 06:30:00 +0000https://yanirseroussi.com/2024/04/29/mentorship-and-the-art-of-actionable-advice/Reflections on what it takes to package expertise and deliver timely, actionable advice outside the context of employee relationships.Assessing a startup's data-to-AI healthhttps://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/Mon, 22 Apr 2024 06:00:00 +0000https://yanirseroussi.com/2024/04/22/assessing-a-startups-data-to-ai-health/Reviewing the areas that should be assessed to determine a startup&rsquo;s opportunities and challenges on the data/AI/ML front.My experience as a Data Tech Lead with Work on Climatehttps://yanirseroussi.com/2024/04/08/my-experience-as-a-data-tech-lead-with-work-on-climate/Mon, 08 Apr 2024 02:00:00 +0000https://yanirseroussi.com/2024/04/08/my-experience-as-a-data-tech-lead-with-work-on-climate/The story of how I joined Work on Climate as a volunteer and became its data tech lead, with lessons applied to consulting &amp; fractional work.Two types of startup data problemshttps://yanirseroussi.com/2024/03/04/two-types-of-startup-data-problems/Mon, 04 Mar 2024 02:00:00 +0000https://yanirseroussi.com/2024/03/04/two-types-of-startup-data-problems/Classifying startups as ML-centric or non-ML is a helpful exercise to uncover the data challenges they&rsquo;re likely to face.Avoiding AI complexity: First, write no codehttps://yanirseroussi.com/2024/02/26/avoiding-ai-complexity-first-write-no-code/Mon, 26 Feb 2024 01:45:00 +0000https://yanirseroussi.com/2024/02/26/avoiding-ai-complexity-first-write-no-code/Two stories of getting AI functionality to production, which demonstrate the risks inherent in custom development versus starting with a no-code approach.Building your startup's minimum viable data stackhttps://yanirseroussi.com/2024/02/19/building-your-startups-minimum-viable-data-stack/Mon, 19 Feb 2024 00:00:00 +0000https://yanirseroussi.com/2024/02/19/building-your-startups-minimum-viable-data-stack/First post in a series on building a minimum viable data stack for startups, introducing key definitions, components, and considerations.Substance over titles: Your first data hire may be a data scientisthttps://yanirseroussi.com/2024/02/05/substance-over-titles-your-first-data-hire-may-be-a-data-scientist/Mon, 05 Feb 2024 02:45:00 +0000https://yanirseroussi.com/2024/02/05/substance-over-titles-your-first-data-hire-may-be-a-data-scientist/Advice for hiring a startup&rsquo;s first data person: match skills to business needs, consider contractors, and get help from data people. \ No newline at end of file