How software teams keep control of codebases that AI now writes most of.
I'm a hands-on architect and engineering leader. Sixteen years across data systems, ML pipelines, and engineering teams, most recently as Director of Engineering at a US digital-health startup, where my teams shipped AI-generated code into a HIPAA and SOC2 production stack every day. Before that, eight years running my own consulting practice.
Agents write the code now. The hard part is everything around that: reviews that cannot keep up, conventions that hold only sometimes, and modules that turn into dark boxes two or three months in. I build the mechanical gates that keep AI-assisted development under control, and I write about what actually works.
Writing
essays, in depthI audited the harness I wrote for a production codebase in 2025: 12,000 lines of rules, zero mechanical checks. What I moved to gates, and what stayed prose.
Consulting
a small number of engagementsWritten application, 15–20 minutes. Personally reviewed within 48 hours. No sales call.
Real artifacts + one working hour. Your setup scored across 20 areas of your AI development system.
A 75-minute walkthrough: where you stand against the frontier and your peers, and a prioritized 90-day plan.
The full shape of the engagement, who it's for, and what the diagnostic covers are on the consulting page.
Start with the application →