Technical PM who builds AI systems.
Documents the AI systems I've built — architecture decisions, what made each solution work, and the cross-system patterns running through all of them: context as files, behavioral contracts, hub document routing, catalog-first conventions, math offloaded to deterministic scripts.
Most AI engineering tutorials are either too shallow for practitioners or benchmark-focused and decay in months. This is a practitioner-depth synthesis built to age well — 14+ docs across prompting, context engineering, agentic systems, evaluation, fine-tuning, and security, structured as a learning path from first principles to production systems.
It doubles as an operational system: a 22-agent catalog, 34 reusable skills, 8 playbooks, and /cook — a 9-phase project scaffold workflow that turns the KB into a decision engine for any new build. Weekly ArXiv digests auto-ingest current research.
- Push notifications: Expanded from one basic alert to a personalized global platform across 4 continents. 140M logins, $2.4M margin impact.
- Cart abandonment: Shipped in 6 weeks. 52% relative conversion lift on projected 7.3M abandoned carts annually.
- Product reviews: Built the business case from the ground up — PRFAQ, revenue model, conversion funnels. Work laid the foundation for a third-party acquisition that would bring the feature to market.
Yale University — B.A. Computer Science, Class of 2023. Ivy League Champion, varsity football.
Oakland, CA · LinkedIn · trenton.johnson@aya.yale.edu