Chelsea Finn

Career

Chelsea Finn completed a PhD in computer science at UC Berkeley in 2018, with the thesis "Learning to Learn with Gradients". Finn's homepage notes the PhD was advised by Pieter Abbeel and Sergey Levine in the Berkeley AI Research Lab.⁵¹

Finn has been an Assistant Professor of Computer Science and Electrical Engineering at Stanford since 2019, and is the William George and Ida Mary Hoover Faculty Fellow. From 2018 to 2024 Finn was a research scientist at Google on the team that began as Google Brain and became part of Google DeepMind, and in 2024 Finn co-founded Physical Intelligence.⁵⁶

Research

Finn's research concerns how robots and other agents can develop broadly capable behavior through learning and interaction, including end-to-end deep learning for vision-based robotic manipulation, meta-learning for few-shot learning, and scaling robot learning to broad datasets. With Pieter Abbeel and Sergey Levine, Finn wrote the 2017 Model-Agnostic Meta-Learning paper.⁶²

Finn's CV lists co-authored papers recognized at robotics and machine learning conferences, including Open X-Embodiment (ICRA 2023 best conference paper), Direct Preference Optimization (NeurIPS 2023 outstanding paper runner-up) and OpenVLA (a CoRL 2024 outstanding paper finalist). At Physical Intelligence, Finn co-authored the 2024 π0 vision-language-action model, released in the openpi repository.⁵³⁴

Teaching

At Stanford, Finn teaches CS224R Deep Reinforcement Learning (Spring 2023, 2025 and 2026) and taught CS330 Deep Multi-Task and Meta Learning each fall from 2019 to 2023, as well as CS221 Artificial Intelligence. Finn has given tutorials on meta-learning at ICML and CVPR and delivered an ICML keynote in July 2024.¹⁵

Work and contributions

• π0 (openpi) Open-source code

  Co-author of "π0: A Vision-Language-Action Flow Model for General Robot Control" (2024). The Apache-2.0 openpi repository, published by the Physical Intelligence team, provides the π0 model with base checkpoints for fine-tuning.³⁴

• Model-Agnostic Meta-Learning (MAML) Published research

  First author, with Pieter Abbeel and Sergey Levine, of the 2017 paper proposing a model-agnostic meta-learning algorithm that trains models to adapt to new tasks from a few examples.²

Timeline

1. 2014Began PhD research in computer science at UC Berkeley.⁵
2. Mar 9, 2017Posted Model-Agnostic Meta-Learning with Pieter Abbeel and Sergey Levine.²
3. 2018Completed the PhD thesis "Learning to Learn with Gradients" and joined Google as a research scientist.⁵
4. 2019Became an Assistant Professor of Computer Science and Electrical Engineering at Stanford University.⁵
5. 2019Taught the first offering of Stanford CS330 Deep Multi-Task and Meta Learning.⁵
6. 2024Co-founded Physical Intelligence.⁵
7. Jul 2024Gave an ICML keynote titled "What Robots Have Taught Me About Machine Learning".⁵
8. Oct 31, 2024Co-authored the π0 vision-language-action flow model paper.³

Other documented roles

Roles the sources document that are not confirmed as current.

• Research Scientist (2018–2024), Google (Google Brain, later Google DeepMind)⁵

Honors and recognition

• Presidential Early Career Award for Scientists and Engineers (PECASE) · United States Government, 2025⁵⁶
• AI2050 Early Career Fellow · Schmidt Sciences, 2024⁵
• NeurIPS Outstanding Paper Runner Up (Direct Preference Optimization) · Conference on Neural Information Processing Systems, 2023⁵
• ICRA Best Conference Paper (Open X-Embodiment) · IEEE International Conference on Robotics and Automation, 2023⁵
• NSF CAREER Award · National Science Foundation, 2023⁵
• Sloan Research Fellowship · Alfred P. Sloan Foundation, 2023⁵⁶
• IEEE RAS Early Academic Career Award in Robotics and Automation · IEEE Robotics and Automation Society, 2022⁵
• ONR Young Investigator Award · Office of Naval Research, 2021⁵
• CoRL Best Paper Award · Conference on Robot Learning, 2020⁵
• ACM Doctoral Dissertation Award · Association for Computing Machinery, 2019⁵⁶
• 35 Innovators Under 35 · MIT Technology Review, 2018⁵

Where they work

• Stanford University 40 people in the wiki
• Physical Intelligence 1 person in the wiki

Related people

• Eric MitchellAlso at Stanford University
• Piero MolinoAlso at Stanford University
• Ian BuckAlso at Stanford University
• Douwe KielaAlso at Stanford University
• Rishi BommasaniAlso at Stanford University
• Misha DenilRobotics · Reinforcement learning
• Sergey LevineRobotics · Reinforcement learning
• Jonathan HurstRobotics · Reinforcement learning
• Justin FuReinforcement learning · Robotics
• Tuomas HaarnojaReinforcement learning · Robotics