Now

July 2026. By day, I consult on AWS for financial-services clients — mostly architecture and delivery work for teams under real regulatory constraints. Nights and weekends, I’m building a self-hosted, mostly-Go lab for AI-agent development, and studying how to make agents reviewable and trustworthy once they’re doing real work: evals that mean something, independent review with real structural distance between the builder and the thing that checks it, and operational patterns for running agents day to day instead of just demoing them. I’m also getting back into conference speaking, after a long gap away from it.