PeopleofSuperintelligence

Researcher

Kamyar Ghasemipour

Also known as Seyed Kamyar Seyed Ghasemipour

Kamyar Ghasemipour is a machine learning researcher who builds robot foundation models for dexterous manipulation as a Founding Member of Technical Staff at Generalist. At the University of Toronto, he conducted undergraduate research in 2014 and 2015 before becoming a graduate student in the Machine Learning Group and the Vector Institute under Rich Zemel. His imitation learning paper on divergence minimization received the Best Paper Award at CoRL 2019, and he published reinforcement learning algorithms at NeurIPS 2019 and ICML in 2021 and 2022. Collaborating with Google Brain, he coauthored the May 2022 Imagen text-to-image paper, which won a NeurIPS 2022 Outstanding Paper Award, and UniSim, which received an ICLR 2024 Outstanding Paper Award. After joining Generalist in May 2024, he coauthored work on self-improving embodied foundation models in September 2025 and helped build the GEN-1 robotics model announced in April 2026.

Now at
Generalist

Timeline

  1. 2014–2015

    Undergraduate research at Toronto

    As a second-year undergraduate at the University of Toronto in summer 2014, he conducts USRA research co-supervised by Sanja Fidler, Suzanne Stevenson and Sven Dickinson. In summer 2015, he works on surface proposals at the University of Toronto through the UTRECS undergraduate research program.

  2. By 2019

    Graduate student in Toronto

    By 2019, he is a graduate student in the Machine Learning Group at the University of Toronto and the Vector Institute, advised by Rich Zemel, with whom he coauthors an ICLR 2018 workshop paper on neural network architecture optimization alongside Will Grathwohl and Elliot Creager. He also works as a long-term student research collaborator with Google Brain Robotics, later Google DeepMind Robotics. In September 2019, he begins an autumn 2019 internship at Google Brain Robotics in Mountain View with Corey Lynch and Pierre Sermanet.

  3. 2019

    Imitation learning and a CoRL award

    In November 2019, he posts "A Divergence Minimization Perspective on Imitation Learning Methods" to arXiv with Richard Zemel and Shixiang Gu, and the paper receives the Best Paper Award at the Conference on Robot Learning (CoRL) in 2019. At NeurIPS 2019, he coauthors "SMILe: Scalable Meta Inverse Reinforcement Learning through Context-Conditional Policies," proposing a maximum-entropy framework that learns policies from few demonstrations.

  4. 2021

    Expected-Max Q-Learning at ICML

    For ICML 2021, he coauthors "EMaQ: Expected-Max Q-Learning Operator for Simple Yet Effective Offline and Online RL" with Dale Schuurmans and Shixiang Shane Gu, introducing the Expected-Max Q-Learning backup operator.

  5. 2022

    Robotic assembly with magnetic blocks

    At ICML 2022, he coauthors "Blocks Assemble! Learning to Assemble with Large-Scale Structured Reinforcement Learning," which introduces a physics-based environment of connectable magnet blocks for robotic assembly.

  6. 2022

    Imagen and a NeurIPS award

    In May 2022, he is one of the fourteen authors of "Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding" from Google Research's Brain Team. The paper presents Imagen, a text-to-image diffusion model that reaches an FID score of 7.27 on the COCO dataset, and introduces the DrawBench benchmark. It wins an Outstanding Paper Award at NeurIPS 2022, which he lists among his December 2022 announcements.

  7. 2023–2024

    UniSim and ALOHA Unleashed

    In October 2023, he coauthors "Learning Interactive Real-World Simulators" with Sherry Yang, Yilun Du, Pieter Abbeel and others; it introduces the UniSim generative simulator and is named an Outstanding Paper at ICLR 2024. In October 2024, he is one of the authors of "ALOHA Unleashed: A Simple Recipe for Robot Dexterity" with Tony Z. Zhao, Chelsea Finn and others, combining ALOHA 2 data collection with diffusion policies for bimanual manipulation.

  8. Since May 2024

    Founding engineer at Generalist

    He joins Generalist in May 2024 as a Founding Member of Technical Staff, where the company's mission is to build robot foundation models for general-purpose dexterous bimanual manipulation. With colleagues at Generalist and Google DeepMind, he coauthors "Self-Improving Embodied Foundation Models" in September 2025, proposing a two-stage post-training approach for robotics. In April 2026, he is part of the Generalist AI research team that announces the GEN-1 robotics model.

Career

  1. Since 2024GeneralistFounding Member of Technical Staff
  2. 2019GoogleStudent Research Collaborator

Education

  1. University of TorontoGraduate student
  2. Vector InstituteGraduate student

From the biography.

  1. Google had the most people shaping AI every year from 2014 to 2025Includes Google. Our estimates of which organizations employed the most people shaping AI each year from 1955 to 2025, tracing the shift from university labs to companies.
  2. Google has the highest combined h-index among AI organizationsIncludes Google. We estimate that Google holds the highest combined h-index at 8,016, while universities account for 73% of the deduplicated total across AI organizations.
  3. 20% of people at companies who shape AI also hold a university role, led by Google and Stanford UniversityIncludes Google. We estimate that 539 of the 6,766 people with a known current organization (8%) hold both a university role and a company role, including 20% of people at companies and 13% of people at universities.
  4. Google alumni have founded or led 133 other organizations, more than any other AI labIncludes Google. We estimate that alumni of 17 major AI labs went on to found or lead 350 other organizations, including 226 they founded, and we trace 472 career moves directly between 23 labs since 2010.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Kamyar Ghasemipour yet. Send us a free-licensed photo.

  1. Sanja Fidler cs.utoronto.ca. Checked 22 October 2026.
  2. Students vclab.ca. Checked 22 October 2026.
  3. Seyed Kamyar Seyed Ghasemipour - University of Toronto cs.toronto.edu. Checked 22 October 2026.
  4. A Divergence Minimization Perspective on Imitation Learning Methods arxiv.org. Checked 22 October 2026.
  5. SMILe: Scalable Meta Inverse Reinforcement Learning through Context-Conditional Policies proceedings.neurips.cc. Checked 22 October 2026.
  6. EMaQ: Expected-Max Q-Learning Operator for Simple Yet Effective Offline and Online RL proceedings.mlr.press. Checked 22 October 2026.
  7. Blocks Assemble! Learning to Assemble with Large-Scale Structured Reinforcement Learning proceedings.mlr.press. Checked 22 October 2026.
  8. Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding arxiv.org. Checked 22 October 2026.
  9. Imagen: Text-to-Image Diffusion Models imagen.research.google. Checked 22 October 2026.
  10. NeurIPS 2022 Awards neurips.cc. Checked 22 October 2026.
  11. Learning Interactive Real-World Simulators arxiv.org. Checked 22 October 2026.
  12. ICLR 2024 Awards iclr.cc. Checked 22 October 2026.
  13. ALOHA Unleashed: A Simple Recipe for Robot Dexterity arxiv.org. Checked 22 October 2026.
  14. [2509.15155] Self-Improving Embodied Foundation Models arxiv.org. Checked 22 October 2026.
  15. Self-Improving Embodied Foundation Models self-improving-efms.github.io. Checked 22 October 2026.
  16. Roboty wchodzą na nowy poziom: GEN-1 osiąga 99% skuteczności i bije rekordy prędkości mlodytechnik.pl. Checked 22 October 2026.