Short thoughts — things I’m building, reading, and chewing on. For longer pieces and my newsletter, see my Substack.

2026
Alignment on the alignment literature — Getting my terms straight on safety, alignment, and control — and why a product eval and a safety eval are asking fundamentally different questions.
2026
Agentic design patterns, read through a healthcare AI lens — Mapping Anthropic's agent patterns onto healthcare — and stumbling into why verifiability is the question that matters.
2026
Colophon: how this site is built — The type and tech behind the design.