creators

Sergey Levine: Berkeley robotics, Physical Intelligence

Sergey Levine is an associate professor at UC Berkeley and a co-founder of Physical Intelligence, working on deep reinforcement learning and robot models.

Sergey Levine is an associate professor of electrical engineering and computer sciences at UC Berkeley, where he runs the RAIL lab, and a co-founder of Physical Intelligence; he states both roles in his own X profile, and Berkeley prints the university title. His own biography records a BS and MS in computer science from Stanford in 2009, a PhD there in 2014, and his arrival on the Berkeley faculty in the autumn of 2016. He works on machine learning for decision making and control: end-to-end training of policies that join perception and control, inverse reinforcement learning, and robots that learn from raw sensory data. He teaches CS 285 and CS 182, both published openly. At Physical Intelligence he is among the authors of the company's robot foundation models: pi-zero, released on 31 October 2024, which folded laundry and bussed tables with one model across several robots, and its successors in November 2025 and April 2026. He received the Presidential Early Career Award for Scientists and Engineers in the 2025 class and writes the Substack newsletter Learning and Control.

Specialization

reinforcement learning, robot learning, robot foundation models, deep learning

Country

United States