Contact

Let's talk & about what your agent fleet is actually doing.

I take on contract and consulting work on agent infrastructure — orchestration, typed memory, trajectory replay, evaluation layers, cost-aware routing. Open to short diagnostic engagements and longer builds. If a system isn't compounding, there's usually a specific reason; finding it is most of the job.

Direct

Send the shape of the problem — current stack, what's breaking or burning budget, what "done" looks like. I'll respond with a read and, if it's a fit, a small scope.

Good first engagement

A useful diagnostic usually starts with one concrete trace: a run that looped, escalated too often, leaked context, or passed review without earning it. I look for the dependency order first — what has to be true before the next ambitious step is safe — then turn that into instrumentation, review gates, and a security-aware operating path.

Channels

Email and GitHub are the current channels. Other surfaces will be added as they become real.

Live demo
/demo/justai