Foundation Models

Build end-to-end machine learning foundation models for robotics control, from data curation and careful evaluation to model architecture and training/inference pipelines. Partner with simulation and real-world robotics teams to create diverse, large-scale datasets for embodied perception and first-person robot video generation. Invent generative simulation techniques to scale and diversify data, and train/evaluate generative models for 3D environments, while helping generate tasks and reward functions.