Scaling Robotics Research
I'm based in SF. Broadly, I'm interested in robot learning, data operations, and inference optimization. I think a lot about the future of labor markets, how personalized healthcare can be scaled, and what ambition looks like post-singularity.
Physical Intelligence | Nov 17, 2025
Our new π*0.6 model can more than double throughput over a base model trained without RL, and can perform real-world tasks: making espresso drinks, folding diverse laundry, and assembling boxes.
Explored sampling efficiency in robotics data.
24-hour coffee shops. U-Dub campus. The Presidio. Linear switches. The Chicago Riverwalk. Vaclav Smil. Cookout Shakes. Ducati. The pencil building on a foggy night. Cheba Hut. Ina Coolbrith Park. The bus-ride from Seattle to Bellevue. Flat whites. The Baroque Era. The NY Times. Brunch. George Hotz's Blog. Donuts!