I build and ship AI products, and I'm moving that work into AI safety. Based in San Francisco, from Buenos Aires.
I spent the last two years building commercial AI products: no-code agents shipped to real businesses, then a self-hosted agent ecosystem I use every day. The question I kept hitting is how an agent behaves once it meets someone who didn't build it. That's what pulled me toward safety.
Currently — exploring AI safety properly in SF, looking for founding or forward-deployed work close to where agents meet real users. Pulling on a thread around evals for mental-health-relevant AI interactions.
Shipped
- 🛒 shop-itinerary — drop in a screenshot, Claude vision extracts the items into a structured shopping itinerary. Next.js 16 · Drizzle · Neon · Vercel AI Gateway. Live
- 🏃 running-plan — a half-marathon plan that keeps itself current, pulling weekly progress live from the Strava API. Vanilla JS · serverless. Live
- 🤖 AI agent ecosystem — self-hosted specialized assistants on OpenClaw with MCP server integrations, configured with Claude Code and Codex.
Before
- 🚀 Co-founded Nubiq — no-code AI agents for SMEs. Ran operations and built the front-end: 3 agents shipped, 5 pilots with real businesses. Awarded the Inspire Up Fund (USD 3,000) by UTDT; finalists in the START Fellowship LATAM.
- 🔍 AI Validator at The Network — manually validated AI-extracted profiles for 2,746 founders across 1,570 companies, correcting outputs below the accuracy threshold.
- 🎓 B.S. in Digital Technology + Minor in Innovation & Entrepreneurship, Universidad Torcuato Di Tella. Exchange semester at ESSEC Business School, France.
Stack — TypeScript · Python · Next.js · React · Tailwind · Postgres · Prisma/Drizzle · Claude · MCP · Vercel
Find me — larabarijhoff.com · LinkedIn · lara.barijhoff1@gmail.com