Codesigning Hardware and VLAs for Manipulation

David Watkins is a Research Lead at the RAI Institute, where he leads foundation model development and large-scale data infrastructure for robot learning. He founded the institute's Foundation Models team and leads the Capture team, a 30-person effort building one of the largest robotic demonstration datasets in the field. His work spans vision foundation models, reinforcement learning from human feedback, and multimodal architectures for embodied AI. 

He won first place at the NeurIPS 2021 MineRL BASALT Competition and has published at CoRL, ICML, IJCAI, and IROS, where his dissertation work was a Best Paper Finalist in Mobile Manipulation. He co-authors the blog "What to Tell the Robot" (whattotelltherobot.com) and is co-authoring Elephants Don't Write Sonnets (MIT Press, 2026) with Stefanie Tellex on the Grounded Turing Test for embodied AI. David holds a Ph.D., M.S., and B.S. from Columbia University, where he was an Army Research Lab Fellow.

A Career-in-progress

A special talk featuring Leo Kim, Director of Engineering at Justworks, as he shares insights from his 30-year journey navigating constant change in the tech industry.

Discover how curiosity, resilience, and continuous professional “refactoring” have shaped a career spanning startups, academia, and mission-critical healthcare technology.