Skip to content
View LobsterQBA's full-sized avatar
1 follower · 1 following

Block or report LobsterQBA

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
LobsterQBA/README.md

Hi, I'm Leo Zhao

I'm a data scientist who builds small AI tools and decision systems. I like work where the reasoning stays visible: what the data says, what the system did, and where a person should still decide.

Recent work

  • Loop Agent — a small agent loop with safe tools, SQLite memory, and a step-by-step trace. It runs locally without an API key.
  • Point2Prompt — point at a UI element and turn feedback into a structured brief for a coding agent.
  • SplitTaste — a reproducible experiment for repairing mixed-household streaming recommendations.
  • Where to Sit — a seat-view simulator for making a better cinema seat choice.

Open-source contributions

  • Microsoft Agent Framework — exposed authorization URLs for Work IQ tool sources and added regression coverage.
  • AO Bench — fixed missing or empty benchmark reports; the change was merged with a passing CI matrix.
  • AO Bench — added a runnable comparison walkthrough for two agent adapters.

I'm interested in agent reliability, useful evaluation, and human-in-the-loop product design.

Personal site

Pinned Loading

  1. loop-agent loop-agent Public

    A small, readable agent loop with safe tools, SQLite memory, and inspectable traces.

    Python

  2. aobench aobench Public

    Forked from MSKazemi/aobench

    Role-aware, permission-enforced benchmark for LLM agents that operate HPC systems — SLURM, telemetry, RBAC, docs, facility. A policy violation hard-fails the task. 88 tasks (10 categories x 5 roles…

    Python

  3. echo-extension echo-extension Public

    JavaScript

  4. trackpad-canvas trackpad-canvas Public

    Your trackpad is the canvas — native macOS architecture sketching.

    Swift

  5. where-to-sit where-to-sit Public

    A Seattle IMAX seat-view simulator for finding where to sit.

    TypeScript

  6. microsoft/agent-framework microsoft/agent-framework Public

    A framework for building, orchestrating and deploying AI agents and multi-agent workflows with support for Python and .NET.

    Python 13.3k 2.2k