
I’ve joined @baseten! Earlier this year, I worked on a data center project and saw how unreliable and fragmented compute supply is. Teams building AI products shouldn’t have to deal with that. Baseten runs across 20+ clouds, makes models faster and cheaper to run, and gives customers one platform to scale on. I care about open-source and customer choice. At my previous startup Llama, we built an open-source smart contract system that our customers could fully own, verify, and build on top of. That same principle applies to AI. Open-weight models allow teams to bring what’s unique to them (their data, workloads, and evals) and post-train a model they actually own. If you’re building an AI product and want to jam, my DMs are (very) open.






