| layout | page | ||||
|---|---|---|---|---|---|
| redirect_from |
|
||||
| comments | false | ||||
| signed | true | ||||
| licence | Creative Commons | ||||
| title | Résumé | ||||
| description | Staff engineer building AI-assisted developer tooling. Python, GCP, LLMs, and infrastructure automation. | ||||
| clippy_help | It looks like you're trying to hire Greg. Consider paying him in cheese 🧀 It's his favourite food! |
Staff engineer building AI-assisted developer tooling and automations.
I ship prototypes quickly and production code elegantly.
Comfortable operating in high pressure remote-first teams.
- Async AI workflow orchestrator (triage -> implement -> assess): built a multi-stage automation system that runs steps asynchronously, preserves context across phases, and uses heuristics to decide when to escalate to a human for guidance.
- AI code reviewer that improves code quality over time: built an internal review system that evaluates changes against human-authored guidelines and uses challenge/verification mechanisms to reduce hallucinations, keeping suggestions actionable and preventing the codebase from "regressing to the mean".
- LLM-powered dbt model optimizer with fail-fast validation: built a tool that proposes SQL optimizations, tests them against sampled datasets to catch issues quickly, and feeds concrete failures back into the LLM to iteratively repair until results reconcile.
- Languages: Python, SQL, Bash, Ruby
- Infra: Linux, Docker, networking, NGINX
- Cloud & Platforms: GCP, AWS, Cloudflare, Tailscale
- Data: dbt, Spark, BigQuery, warehouses at scale
- AI/LLM systems: LLM benchmarking, context, tool workflows, guardrails
- Serving & runtimes: vLLM, Ollama, llama.cpp
- Ops & delivery: CI/CD, observability
- Microcontrollers: ESP32s, Raspberry Pis, PCBs
- Built an LLM-powered dbt optimization tool that fails fast on sampled data, validates semantic equivalence, and iterates automatically by feeding execution/validation failures back to the model.
- Improved reliability by adding validation gates for AI-generated changes (language-aware checks, clearer failure states, and safer automation defaults).
- Built an async AI automation orchestrator for data engineering workflows using a staged approach (triage -> implement -> assess) with human-in-the-loop escalation driven by heuristics and confidence checks.
- Built an AI code review system that reviews changes against human-authored engineering guidelines, using challenge/verification techniques to reduce hallucinations and making suggestions one-click apply in GitHub.
- Migrated core warehouse workloads from Spark to dbt on GCP.
- Improved modeling practices, review workflows, and operational consistency across a large shared codebase.
- Led a small team delivering forecasting and infrastructure planning recommendations.
- Developed tooling to make load balance recommendations for our core infrastructure.
- Learned how to make a small human happy.
- Built a personal analytics product end-to-end (backend, frontend, infra, and ops).
- Worked across Rails, Python, GraphQL, React, Flutter, and JavaScript on cloud + self-hosted infrastructure.
- Site: https://memair.com
- Sailing SV Catsaway from Kingston, Ontario, to the Bahamas and back.
- Produced a YouTube series documenting our adventure, honing video communication skills.
- Hacking on a few (mostly boat-related) projects.
- Data modeling, reporting, experimentation, and ML support for early Shopify.
- Worked primarily in Python/Ruby with Spark, Presto, and Rails.
- Supporter database projects; Rails + SQL; operational reporting and process improvement.
- Peacemaking tour in Afghanistan 🇦🇫.
- Peacekeeping tour in Solomon Islands 🇸🇧.
- Teaching Acroyoga as part of SmileyOm.
- Home lab: servers, Docker workloads, monitoring, backups, and secure remote access.
- Self-hosting (incl. Mastodon) and home automation (Home Assistant / ESP32).
- Speaker: AI Tinkerers Toronto (Nov 2025) and many internal Shopify communications.
- Blog posts.
Bachelor’s Degree - Murdoch University (2008 – 2011, Remote)
Focus: security / terrorism / counter-terrorism
- Canada: Enhanced Reliability (valid through 2029)
- Australia: Top Secret (lapsed)
🇦🇺 Australian
🇨🇦 Canadian
🇮🇪 Irish (🇪🇺EU)
🇬🇧 British
- Improve the world according to my Codex Vitae.
- Use AI & data to solve interesting problems.
- Lead great teams; keep learning; avoid being the smartest person in the room.
Connection:
- 3Gbps ⬆️⬇️ fiber Internet
- Ubiquiti networking equipment with ethernet to office
- 5G cellular & 1hr UPS redundancy
Residence: Prince Edward County, Ontario, Canada
- 🏢 10mins to co-working spaces
- 🚗 15mins to 401 highway
- 🚊 15mins to Belleville Via Rail station
✈️ 2hrs to Toronto Pearson International Airport
Timezone: Eastern Standard Time (UTC−05:00)
Workable hours:
- PST: 04:00 - 13:30 PST
- EST: 07:00 - 16:30 EST
- UTC: 12:00 - 21:30 UTC