Evidence-backed roles, scope, and outcomes. Focus on taste, product intuition, and AI collaboration over pure execution.
Led 8-person team owning the core AI platform (models, tooling, evals) for 50M+ MAU product. $2M annual model budget. Shipped 3 major model upgrades, 40% latency reduction, 25% cost reduction. Outcome: Launched in-house eval framework adopted org-wide; reduced hallucination rate from 12% → 3% on critical paths; built PM-facing dashboard for model quality tracking.
Owned developer experience for 500k+ developers. 12-person team across CLI, SDKs, docs, and API platform. $1.5M budget. Shipped v2 API, new CLI, interactive tutorials. Outcome: API v2 adoption 80% in 6 months; CLI downloads 3x; 'Time to First Hello World' from 45 min → 3 min; NPS +34.
Model selection, eval frameworks, routing, cost/quality tradeoffs, PM-facing quality dashboards
API design, CLI/SDK architecture, onboarding funnels, interactive tutorials, time-to-value optimization
Agent UX patterns, review/approval loops, trust calibration, hallucination mitigation, context management
Teardown methodology, magic vs. friction analysis, concrete PM recommendations, evidence-backed decisions