About
I’ve spent six years shipping production systems, including ML with real stakes — a COVID diagnostics ensemble that read PCR curves in the browser with zero reported errors across a million-plus diagnoses. Lately it’s been RL post-training, verifiable evaluation, and the tooling that makes model behavior something you measure instead of guess. The map above has all of it.
Currently interested in AI and developer tooling, full-stack product engineering, and research-engineering problems.