Independent engineer building dependable AI systems and developer infrastructure.
I build software for places where “it seemed to work” is not enough: AI evaluation, release evidence, observability, developer tooling, and trust boundaries around agent capabilities.
I like narrow products with explicit responsibilities. The system should say what it knows, what it cannot establish, and what a human still needs to decide. That usually produces calmer software—and much better operational decisions.
This site is where I document that work, publish independent products, and make it easier for teams with a difficult systems problem to find me.