Now

Building

I'm heads-down on SkillCreator.ai, a local- first desktop IDE for writing and evaluating AI agent skills. The evaluation framework runs across 8 agent runtimes and scores each skill on six dimensions: correctness, tool selection, instruction adherence, latency, token efficiency, and error recovery. The CLI companion for installing those skills, Ai-Agent-Skills, is at 1K+ stars and 12K+ npm installs and still getting commits every week.

I'm also keeping Halaali shipping. The halal-discovery app has 500+ users and 1,300+ indexed locations, and every release means a production mobile deploy, a PostGIS migration, and whatever else Stripe throws at me. Good forcing function.

Reading and learning

Agent evaluation is a mess right now. Everyone has opinions about how to measure a skill; nobody agrees on the metric. I'm working through the eval frameworks that already exist and figuring out which parts I can adopt for SkillCreator and which parts I have to invent.

Also following the Anthropic and Vercel engineering posts, plus whatever new tool patterns ship in the MCP protocol spec.

Looking for

A full-time role at a small team building agent tooling, developer tools, or 0-to-1 AI products. Remote, or Toronto in person. Founding engineer, founding product engineer, or early product engineer is the shape. Small team, fast shipping, real users.

Canadian citizen, TN visa eligible for US, available immediately. If you're hiring for something like that, email me.

Personal

Based in Toronto. 23 years old. Canadian citizen, family close by. Shipping software since I was 17.