# Pete Florence

> Pete Florence is the co-founder and CEO of Generalist, which develops embodied foundation models for robots. A former Google DeepMind senior research scientist and MIT computer science PhD, he has co-authored research including Dense Object Nets, Transporter Networks, Implicit Behavioral Cloning, PaLM-E and RT-2. He also co-founded Stage One Education with Andrew Barry, developing hands-on engineering workshops for students.

- **Role:** Co-founder and CEO at Generalist
- **Organizations:** Generalist, Google DeepMind, Google, Massachusetts Institute of Technology, Stage One Education
- **Education:** Undergraduate degree in Chemistry, Princeton University, Master’s in Physics, University of Cambridge, PhD in Computer Science; advisor Russ Tedrake, Massachusetts Institute of Technology
- **Known for:** Co-founder and CEO of Generalist, building embodied foundation models for physical interaction., Co-author of Dense Object Nets, winner of the CoRL 2018 best paper award., Co-author of Self-Supervised Correspondence in Visuomotor Policy Learning, listed among IEEE RA-L paper award winners.

## Career timeline

- **2012** — Graduated from Princeton, where he studied chemistry; subsequently completed a one-year physics master’s at Cambridge.
- **2014** — Began doctoral study at MIT with Russ Tedrake.
- **2018** — Co-authored NanoMap and Dense Object Nets; Dense Object Nets received the CoRL best paper award.
- **2019** — Completed the MIT thesis Dense Visual Learning for Robot Manipulation.
- **2020** — Co-authored Transporter Networks; Self-Supervised Correspondence in Visuomotor Policy Learning appeared in RA-L volume 5 and received an RA-L paper award.
- **2021** — Led the author list of Implicit Behavioral Cloning, presented at CoRL.
- **2023** — Co-authored PaLM-E and RT-2 while working in Google robotics research.
- **2025-11** — Generalist introduced GEN-0, reporting scaling behavior in embodied foundation models.
- **2026-04** — Generalist introduced GEN-1; Florence co-authored an essay explaining the goal-driven research approach.
- **2026-06** — Generalist announced $400 million in new funding, taking total funding above $500 million.
- **2026-08** — Generalist introduced GEN-1.5, reporting one-shot and few-shot adaptation on short physical tasks.

## Achievements

- Co-founder and CEO of Generalist, building embodied foundation models for physical interaction.
- Co-author of Dense Object Nets, winner of the CoRL 2018 best paper award.
- Co-author of Self-Supervised Correspondence in Visuomotor Policy Learning, listed among IEEE RA-L paper award winners.
- Co-author of DeepSDF, a CVPR 2019 best paper finalist.
- Co-author of Transporter Networks, Implicit Behavioral Cloning, PaLM-E and RT-2.
- Co-founded Stage One Education; his personal website reports more than 10,000 workshop students.

## Latest updates

- **2026-10** — Appeared in Mario Gabriele’s October 20 interview, When Robots Improvise, discussing dexterity, learning, physical data collection and leadership.
- **2026-08** — Generalist released GEN-1.5 on August 19. It reported 59% average success with one-shot in-context prompting across 10 short tasks and 83% after 10 gradient steps using 5 minutes of task data.
- **2026-07** — Generalist reported GEN-1 transfer across a wider range of robot hands and specialized end effectors.
- **2026-06** — Generalist announced $400 million in new funding led by Radical Ventures, with total funding above half a billion dollars.

## Links

- Website: https://www.peteflorence.com/
- LinkedIn: https://www.linkedin.com/in/peteflorence
- Twitter/X: https://x.com/peteflorence
- GitHub: https://github.com/peteflorence

---

Profile page: https://yespress.io/pete-florence
Published by YesPress — https://yespress.io
Last updated: 2026-10-22