About
The short version, in order.
- 2025 –Zapier — Research engineer on agent evaluation: simulated worlds, eval-driven development, release gates for an agent platform used by 1.5M+ people.
- 2022 – 25Taltrics — Co-founder and CTO. Resource forecasting for enterprise operations, then a multi-agent platform that scopes and prices professional-services deals on the call; Berkeley SkyDeck; FABA top-3 French entrepreneur in the US, 2024.
- 2020 – 23Z-Inspection — Researcher in an international group assessing trustworthy AI in healthcare; four peer-reviewed papers.
- 2019 – 22Genentech / Roche — Senior data scientist. Discrete-event simulation of a 300-trial portfolio; planning and crisis-response models used across therapeutic areas.
- 2018 – 21AI for Good Foundation — Led the UN SDG Data Catalog; Microsoft funding; SIGKDD 2019.
- 2018 – 19UC Berkeley — MEng, Industrial Engineering & Operations Research. Played soccer for Cal.
- – 2018École Centrale Paris — Applied mathematics and computer science.
- Feb 2026Zapier × Gamma
- Sep 2024Socially Conscious AI Podcast
- Jun 2022
- Jul 2020
- Aug 2019SDG Data Catalog — Social Impact ShowcaseSIGKDD 2019
Running has always been my reset — middle distance, 800 m and 1500 m, and I still chase small gains that only show up on race day. I played soccer for years, including at Berkeley. These days I'm outside when I can: hiking, skiing in the Alps when I visit family, biking around Marin, learning to surf.
Books I keep returning to: Taleb'sIncerto, Pearl'sThe Book of Why, Kahneman'sThinking, Fast and Slow, O'Neil'sWeapons of Math Destruction. Podcasts:Dwarkesh,Lex Fridman,Generation Do It Yourself.
Set in Newsreader, with Inter and JetBrains Mono for structural indices and diagrams. Built with Astro to static HTML with inline SVG simulations and zero external runtime dependencies. No tracking. Source onGitHub.
andyspezzatti@gmail.com — I read everything, and reply to most of it. I occasionally advise teams on agent evaluation; email is the way.