About

I work on applied AI. Mostly that means taking models from “impressive demo” to “reliably useful”: post-training LLMs, using reinforcement learning where it actually fits, and building the evaluations that tell you whether any of it worked.

Before that I worked across engineering, platform and automation.

This blog is where I write things up after trying them: experiments, comparisons, architecture decisions, and the ideas that didn’t pan out. Opinions are my own.