Universal Mechanical Assistant

Pioneer of real-world deep learning at scale: shipped Tesla Autopilot to millions of cars, developed Tesla Optimus AI from scratch, and democratized robot learning by founding LeRobot at Hugging Face.

His mission: building general-purpose robots to address the important challenges of our time.

He has spent 25+ years pushing the boundaries of design, manufacturing, and rapid robot development in humanoid robotics.

An original pioneer of anatomically accurate robots with CRONOS and ECCERobot, and more recently designer of the world’s most popular 3D-printed robot arm, the open-source SO-100.

He brings over two decades of AI and robotics research at NYU, Google Brain and DeepMind, earning 100,000+ citations and winning top computer vision competitions.

As a founding member of Google DeepMind’s robotics team, he developed the first manipulation VLA and advanced embodied safety, making him one of the field’s most complete researchers.

As co-founder of LeRobot at Hugging Face, he built the project’s codebase and infrastructure, driving its rise into one of the most impactful AI-for-robotics platforms.

His work on data formats and middleware are now widely used across industry and research.

Founding member of the Multimodal team at Hugging Face, Founding-team Research Engineer at H Company.

PhD in robust reinforcement learning, architect of LeRobot’s RL system reaching state-of-the-art robot learning speed, and core team member of River ML: the online ML library with 2M+ downloads.

He has worked on scene understanding and forecasting for autonomous vehicles and large-scale VLMs for agentic AI. PhD on Generative Models (2020).

Large-scale training. Efficiency. Multimodal. VLAs

After two decades at Citigroup investment bank, he leverages deep structuring and scaling expertise to lead UMA’s finance and strategy.

An industrial engineer turned entrepreneur, she brings practical skills and strong execution, and recently led the LeRobot Worldwide Hackathon at Hugging Face.