PeopleofSuperintelligence

Researcher

Konrad Żołna

Konrad Żołna is a machine-learning researcher whose work spans sequence models, imitation learning, generalist agents and world models. After winning a third prize at the International Mathematics Competition in 2012, studying wind-turbine condition monitoring in 2015 and presenting the LSTM system user2vec at ICML in June 2016, he earned a computer science doctorate at Jagiellonian University in July 2021 while publishing on focused hierarchical RNNs, adversarial imitation learning and offline reinforcement learning. At Google DeepMind, where he served as a Senior Research Scientist from 2022 to 2024, he coauthored the generalist agents Gato in May 2022 and RoboCat in June 2023, as well as the world model Genie, which won an ICML Best Paper award in 2024. Since 2024, he is a Member of Technical Staff at Microsoft AI, where he works on reinforcement learning and robotic manipulation.

Now at
Microsoft
Badges

Timeline

  1. 2012

    Mathematics competition prize

    Representing Jagiellonian University in 2012, he wins a third prize with 29 points at the International Mathematics Competition for University Students.

  2. 2015

    Wind-turbine condition monitoring

    In 2015, he coauthors Nonlinear Cointegration Approach for Condition Monitoring of Wind Turbines in Mathematical Problems in Engineering.

  3. 2016

    User modeling with LSTMs

    He presents user2vec, an LSTM approach to website user modeling, at ICML's Online Advertising Systems Workshop in New York in June 2016.

  4. 2017–2021

    Doctoral research and sequence models

    At Jagiellonian University, he pursues doctoral research in the Department of Machine Learning from 2017 to 2021, receiving a computer science doctorate in July 2021 for Robustness of Neural Networks under supervisor Grzegorz Świrszcz of DeepMind after a proceeding opened in April 2017. At ICML in June 2018, he coauthors Focused Hierarchical RNNs for Conditional Sequence Processing, which reads one token at a time and selects information relevant to its context or question. He also coauthors Task-Relevant Adversarial Imitation Learning in October 2019 for robot imitation rewards, Offline Learning from Demonstrations and Unlabeled Experience in November 2020 for offline reinforcement learning, and Robust Learning-Augmented Caching: An Experimental Study at ICML in 2021.

  5. 2022

    The Gato generalist agent

    In May 2022, he coauthors Gato, a multimodal generalist policy whose same network and weights play Atari, caption images, chat and control a real robot arm to stack blocks.

  6. 2023

    RoboCat for robotic manipulation

    In June 2023, he coauthors RoboCat, a generalist agent for robotic manipulation that adapts to new tasks and robots with 100 to 1,000 examples. The model also generates data for later training rounds.

  7. 2024

    Genie wins Best Paper

    In February 2024, he coauthors Genie, an 11-billion-parameter world model trained on unlabeled internet videos to generate action-controllable virtual worlds from text prompts, images, photographs and sketches. ICML gives Genie a Best Paper award in 2024.

  8. Since 2024

    Microsoft AI technical staff

    He joins Microsoft AI as a Member of Technical Staff in 2024, after working as a Senior Research Scientist at Google DeepMind from 2022 to 2024, with research focused on reinforcement learning and robotic manipulation. In January 2025, he gives a GMUM seminar titled Towards Generalist Agents: A Dive into Gato, RoboCat, GATS, and Genie. At ICLR in April 2025, he joins Yazhe Li for Microsoft AI's recruiting Q&A on Microsoft Search, Assistant and Intelligence.

Career

  1. Since 2024MicrosoftMember of Technical Staff
  2. 2022–2024Google DeepMindSenior Research Scientist

Education

  1. 2021Jagiellonian UniversityPhD

From the biography.

  1. Of 943 people shaping AI who worked at Google, 446 work there nowIncludes Konrad Żołna. Our estimates of where the 943 people shaping AI who worked at Google and Google DeepMind work now: 446 at Google and 452 across 278 other organizations.
  2. Google alumni have founded or led 133 other organizations, more than any other AI labIncludes Konrad Żołna. 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.
  3. 16% of AI career moves in the 2020s went from a university to a companyIncludes Konrad Żołna. We estimate that 16% of AI career moves in the 2020s went from a university to a company; 29% of moves starting at a university went to a company, and 76% ended the university role.
  4. The median h-index is 48 at universities and 14 at frontier AI labsIncludes Konrad Żołna. The median h-index is 48 at universities and 14 at frontier AI labs. We compare publishing careers, recent citations, dual appointments and job moves across 2,420 university researchers and 308 frontier-lab researchers.

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 Konrad Żołna yet. Send us a free-licensed photo.

  1. XIX Międzynarodowe Zawody Matematyczne mimuw.edu.pl. Checked 22 October 2026.
  2. dr Konrad Żołna apacz.matinf.uj.edu.pl. Checked 22 October 2026.
  3. Our user2vec at ICML 2016 techblog.rtbhouse.com. Checked 22 October 2026.
  4. Focused Hierarchical RNNs for Conditional Sequence Processing - Microsoft Research microsoft.com. Checked 22 October 2026.
  5. [1910.01077] Task-Relevant Adversarial Imitation Learning arxiv.org. Checked 22 October 2026.
  6. [2011.13885] Offline Learning from Demonstrations and Unlabeled Experience arxiv.org. Checked 22 October 2026.
  7. Wydział Matematyki i Informatyki - View Procedure apacz.matinf.uj.edu.pl. Checked 22 October 2026.
  8. Obrony prac doktorskich - Wydział Matematyki i Informatyki - Faculty of Mathematics and Computer Science - Jagiellonian University matinf.uj.edu.pl. Checked 22 October 2026.
  9. Robust Learning-Augmented Caching: An Experimental Study proceedings.mlr.press. Checked 22 October 2026.
  10. [2205.06175] A Generalist Agent arxiv.org. Checked 22 October 2026.
  11. [2306.11706] RoboCat: A Self-Improving Generalist Agent for Robotic Manipulation arxiv.org. Checked 22 October 2026.
  12. Genie: Generative Interactive Environments - Google DeepMind deepmind.google. Checked 22 October 2026.
  13. ICML 2024 Awards icml.cc. Checked 22 October 2026.
  14. Konrad Zolna - Lacuna lacuna.tiptreesystems.com. Checked 22 October 2026.
  15. Seminars | GMUM gmum.net. Checked 22 October 2026.
  16. Microsoft at ICLR 2025 - Microsoft Research microsoft.com. Checked 22 October 2026.