A stateless search architecture that eliminates the vector database entirely, compliance features designed as architectural patterns instead of checkboxes, and managing AI development context as production infrastructure.
Product Manager · 9 years at PayPal · Now building AI products
I believe the system prompt is the product spec — and that every feature you don't build is a risk you don't carry.
I spent nine years at PayPal building enterprise platforms that served 90,000+ employees & contingent workers globally — onboarding systems, internal tools, automation that cut support tickets by 50%. Now I'm applying that product discipline to AI: building a consumer AI companion app from scratch with multi-layer safety architecture, adversarial testing, and compliance design for three emerging AI laws.
Based in the Bay Area. Open to full-time and fractional AI PM roles.
A mobile-first PWA for nightly gratitude journaling, powered by an AI companion named Luna. Built solo from architecture to deployment.
Problem
AI companion apps are shipping without safety infrastructure — no crisis protocols, no adversarial testing, no compliance with emerging AI legislation.
Approach
Safety-first architecture where every feature is evaluated against a 4-tier prioritization framework: safety/legal → core functionality → user trust → growth.
Key Decisions
Stack
React · Tailwind CSS · Supabase (PostgreSQL + Edge Functions) · Gemini 2.5 Flash · Vercel
Building working memory into an AI companion, letting the model author its own UX signals, and why a 429 error in a journaling app is an emotional rupture — not a technical inconvenience.
Why crisis detection must bypass the LLM entirely, how adversarial testing exposed the cost of false positives, and building an AI personality framework where tone is risk management — not cosmetic.
Seven security layers mapped to OWASP LLM Top 10, a system prompt rewritten as a behavioral spec, and the discovery that your safety infrastructure can fight itself at the worst possible moment.
Filter by keyword or select a tag.
A stateless search architecture that eliminates the vector database entirely, compliance features designed as architectural patterns instead of checkboxes, and managing AI development context as production infrastructure.
Migrating off a no-code platform mid-build, designing a summarization feature as a side-channel architecture, and the debugging tax that comes with every AI-generated line of code.