PeopleofSuperintelligence

Researcher

Nikita Zhivotovskiy

Nikita Zhivotovskiy is an assistant professor in UC Berkeley’s Department of Statistics who studies statistical learning theory and online learning. He defended his thesis at the Moscow Institute of Physics and Technology in 2018, then held postdoctoral research posts at Google Research in Zürich and ETH Zürich. With Olivier Bousquet, Steve Hanneke and Shay Moran, he coauthored a paper on learning guarantees for support vector machines that won the 2020 COLT Best Paper Award. His later work examined learning across multiple data distributions and efficient methods for online logistic regression.

Works
61
Citations
359
h-index
11
Badges

Publication numbers from OpenAlex, as of 28 October 2026.

Timeline

  1. 2018

    A thesis at MIPT

    He defends his thesis at the Moscow Institute of Physics and Technology in 2018 under Vladimir Spokoiny and Konstantin Vorontsov. During his time in Moscow, he holds part-time affiliations with the Institute for Information Transmission Problems, Higher School of Economics and Skoltech. Before his January 2019 Google appointment, he also spends half a year in the Department of Mathematics at Technion, hosted by Shahar Mendelson.

  2. January 2019–December 2020

    Research at Google Zürich

    From January 2019 to December 2020, he works as a postdoctoral researcher at Google Research in Zürich, hosted by Olivier Bousquet.

  3. 2020

    A COLT Best Paper Award

    With Olivier Bousquet, Steve Hanneke and Shay Moran, he coauthors “Proper Learning, Helly Number, and an Optimal SVM Bound” in 2020. The paper gives an optimal bound on the number of training examples needed by support vector machines in the realizable setting, and wins the 2020 Conference on Learning Theory Best Paper Award.

  4. January 2021–October 2022

    Postdoctoral research at ETH

    From January 2021 to October 2022, he works as a postdoctoral researcher in ETH Zürich’s Department of Mathematics, hosted by Afonso Bandeira.

  5. 2023

    Optimal PAC learning bounds

    In 2023, he coauthors “Optimal PAC Bounds Without Uniform Convergence” with Ishaq Aden-Ali, Yeshwanth Cherapanamjeri and Abhishek Shetty. The work turns leave-one-out prediction errors into high-probability risk bounds and gives optimal guarantees for binary and multiclass classification.

  6. 2024

    Learning across multiple distributions

    With Kasper Green Larsen and Omar Montasser, he coauthors “Derandomizing Multi-Distribution Learning” in 2024. The paper studies a single predictor trained to work across several data distributions, proves that removing randomness is computationally hard in general, and identifies a structural condition that permits efficient conversion to deterministic predictors.

  7. 2025

    Online classification with relaxed benchmarks

    In 2025, he coauthors “Beyond Worst-Case Online Classification: VC-Based Regret Bounds for Relaxed Benchmarks” with Omar Montasser and Abhishek Shetty at the Conference on Learning Theory. The algorithms compare predictions with benchmarks that allow input perturbations, Gaussian smoothing or a prescribed output margin.

  8. April 2026

    Efficient online logistic regression

    In April 2026, he coauthors “Efficient Logistic Regression with Mixture of Sigmoids” with Federico Di Gennaro and Saptarshi Chakraborty. The paper gives an efficient implementation of Exponential Weights with a Gaussian prior for online logistic regression while retaining its near-optimal worst-case regret guarantee.

  9. As of October 2026

    Assistant professor at Berkeley

    As of October 2026, he is an assistant professor in UC Berkeley’s Department of Statistics, with research interests in statistical learning theory, online learning, nonparametric estimation, hypothesis testing and applied probability. In fall 2026, he teaches Introduction to Probability at an Advanced Level.

Career

  1. NowUniversity of California, BerkeleyAssistant Professor
  2. 2021–2022ETH ZurichPostdoctoral Researcher
  3. 2019–2020Google ResearchPostdoctoral Researcher

Education

  1. 2018Moscow Institute of Physics and TechnologyThesis defense

From the biography and OpenAlex affiliations (the years of papers that list the organization), so some years are approximate.

Checked career sources: isye.gatech.edu.

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 Nikita Zhivotovskiy yet. Send us a free-licensed photo.

  1. ISyE Seminar Speaker - Nikita Zhivotovskiy | H. Milton Stewart School of Industrial and Systems Engineering isye.gatech.edu. Checked 28 October 2026.
  2. Nikita Zhivotovskiy | Simons Institute simons.berkeley.edu. Checked 28 October 2026.
  3. Nikita Zhivotovskiy sites.google.com. Checked 28 October 2026.
  4. Proper Learning, Helly Number, and an Optimal SVM Bound proceedings.mlr.press. Checked 28 October 2026.
  5. Optimal PAC Bounds Without Uniform Convergence arxiv.org. Checked 28 October 2026.
  6. Derandomizing Multi-Distribution Learning arxiv.org. Checked 28 October 2026.
  7. Beyond Worst-Case Online Classification: VC-Based Regret Bounds for Relaxed Benchmarks proceedings.mlr.press. Checked 28 October 2026.
  8. Efficient Logistic Regression with Mixture of Sigmoids arxiv.org. Checked 28 October 2026.
  9. Nikita Zhivotovskiy | Department of Statistics statistics.berkeley.edu. Checked 28 October 2026.
  10. Nikita Zhivotovskiy | Research UC Berkeley vcresearch.berkeley.edu. Checked 28 October 2026.