Chelsea Finn
Chelsea Finn is a Stanford assistant professor and co-founder of Physical Intelligence. Her research asks how learned systems can adapt to new tasks and act in changing physical environments. She studied at MIT and completed her doctorate at Berkeley. With Ian Goodfellow and Sergey Levine, she worked on predicting how robot actions change images. Finn, Pieter Abbeel, and Levine then introduced Model-Agnostic Meta-Learning, a method for adapting across tasks with a few further updates. Her later shared papers include ALOHA, which learns precise two-handed actions from demonstrations, and Direct Preference Optimization for language-model training. At Stanford, she leads the IRIS lab and teaches learning methods. Her company role continues alongside academic research and teaching, with product performance requiring its own evidence.
- Now at
- Stanford University and Physical Intelligence
- Works
- 360
Publication numbers from OpenAlex, as of 14 October 2026. Citations and h-index are hidden because this OpenAlex record may be incomplete or mix in other authors.
Timeline
-
University study
MIT and Berkeley
Finn studies electrical engineering and computer science at MIT, then completes a Berkeley computer science doctorate.
-
2016
Predicting interaction
Finn, Ian Goodfellow, and Sergey Levine study action-conditioned video prediction for physical tasks.
-
2017
Meta-learning
Finn, Pieter Abbeel, and Levine introduce Model-Agnostic Meta-Learning.
-
Stanford
IRIS research and teaching
Her lab studies intelligence through robotic interaction and adaptation across tasks.
-
2023
Robots and preferences
She coauthors the ALOHA fine-manipulation research and Direct Preference Optimization with separate teams.
-
October 2026
University and company work
She is a co-founder of Physical Intelligence and continues to teach at Stanford.
Sources and credits
Photo: UC Berkeley, via Wikimedia Commons. License: CC BY 3.0. Source: commons.wikimedia.org. Cropped to a square and resized. Ask us to remove or replace this photo.
