Resources

Resources

Things I keep coming back to, plus notes I wrote because the reference I wanted didn't exist.

Posts

Stop taking AI advice on where to submit your paper

I asked GPT-6 Luna where to submit 132 blinded AI papers whose 2026 venues it could not have seen. It never suggested an oral, sent half the workshop papers to main conferences, and sounded just as sure when wrong.

October 6, 2026

Agents

Claude API documentation

Authoritative reference for the Messages API, tool use, prompt caching, and the agent surfaces built on top of them.

August 1, 2026
Railtracks: an open-source agentic framework

The framework I designed at Railtown for building, debugging and monitoring agentic systems. Canada's first open agentic framework.

September 1, 2025

LLM Ops

What AEval measures, and why

Why accuracy alone does not tell you whether an expert system is safe to ship, and why hallucination, security and fairness each need a different harness.

June 15, 2026