Cibi is
I’ve spent eight years building new products. First as an engineer at Faire, then as a PM at ADP and BLD.ai, then as a founder.
This year I went full-time on AI safety. I had more than 300 conversations with researchers, founders and grantmakers about where I could help. One problem kept coming up. AI can now run alignment experiments on its own, and checking which results to trust is still mostly done by hand.
That’s what I’m working on. I’m not a career ML researcher, so I started with what I know how to do. I found a researcher with this problem and built a demo for him.
Now
- Crux
- An early concept I’m developing for checking alignment research done by AI. A second model tries to prove the first one wrong, and the researcher only sees what survives. I built the first demo at BlueDot’s AI for Epistemics hackathon, with an incoming alignment scientist at AE Studio as my design partner. Read the announcementSee the deck ↗
- Open to
- Alignment research roles and fellowships. If you work on oversight, I’d like to hear from you. Email me.
The path so far
- Engineer2017–19
- Started as a mobile engineer. Shipped apps for Carnival Cruise Line and CBS Radio, then joined Faire early as its third iOS engineer, working on search and checkout.
- Product2020–24
- Moved into product. At ADP in New York I took a low-code platform to its first enterprise customers. At BLD.ai I went from PM to Lead PM. I grew an AI meeting assistant to 70K monthly users, ran a performance and pay platform used by Uber, Coinbase and Two Sigma, and led a venture studio that launched seven products. One was acquired, and one partnered with SFMOMA.
- Founder2024–26
- Started Nirvana, building tools for focus and wellbeing. Tested more than 30 problem ideas and got 100+ sign-ups, including people at OpenAI, Meta and Coinbase, plus enterprise pilot offers.
- AI safety2026–
- Full-time. Building Crux and writing here.
Computer Science, University of Waterloo · LinkedIn ↗