About

I am a developer who builds agentic systems and the infrastructure around them. The work on this site runs from orchestrating a local fleet of AI agents on my own machine, to wrapping autonomous agents in a security perimeter, to turning piles of documents into a queryable knowledge base, to taking sixty-year-old languages like COBOL and Apollo-era assembly and flying them with a modern model.

A few instincts show up in nearly all of it. I design database-first, so that correctness is a property of the data rather than a matter of constant vigilance. I default to local-first, spending the expensive cloud model only where it genuinely earns its place. And I tend to treat the system itself as the thing being designed, giving every part one clear job and putting an intelligent orchestrator at the center.

I care a great deal about doing this safely. Several of these projects are, at heart, about giving a capable model real agency inside constraints it cannot override, which I think is one of the central problems of building with AI well. This site is the show-rather-than-tell version of that: the work is here to be read, run where possible, and judged on its own terms.