PeopleofSuperintelligence

Researcher

Grzegorz Swirszcz

Also known as Grzegorz Świrszcz

Grzegorz Swirszcz is a mathematician and machine-learning researcher who coauthored AlphaTensor, the DeepMind system that used reinforcement learning to discover faster matrix multiplication algorithms. Trained in dynamical systems, he earned a PhD at Warsaw University in 1998 and joined IBM's Watson Research Center in 2004, where his work shifted toward modeling and machine learning. In 2009, he coauthored the winning solution for the KDD Cup Orange Challenge, and from 2016 he worked on AI at DeepMind, coauthoring papers on local minima in neural network training and on Sobolev training. With nine coauthors, he presented an AI system in January 2025 that generates complex polytopes for problems such as the Hirsch conjecture. He is Quant Research Director at the London hedge fund Qube Research & Technologies, which he joined in September 2024.

Timeline

  1. 1998

    A Warsaw PhD in mathematics

    He earns master's and doctoral degrees in the Department of Mathematics at Warsaw University, completing a PhD in 1998 on limit cycles of polynomial vector fields under advisor Henryk Żołądek. After the PhD, he continues research on dynamical systems and ordinary differential equations as a postdoc at the Centre de Recerca Matemàtica in Barcelona, work that later forms his habilitation from the Institute of Mathematics of the Polish Academy of Sciences.

  2. 2004

    Joining IBM T.J. Watson Research

    In 2004, he joins IBM T.J. Watson Research, where his work shifts toward applications of mathematics in modeling and machine learning, including financial modelling.

  3. 2009

    Winning the KDD Cup Orange Challenge

    In 2009, he coauthors the winning solution for the KDD Cup Orange Challenge, an ensemble-selection approach, with Alexandru Niculescu-Mizil, Claudia Perlich and nine other researchers.

  4. 2016

    Joining DeepMind

    He begins working on artificial intelligence at DeepMind in 2016. In November 2016, he, Wojciech Marian Czarnecki and Razvan Pascanu post "Local minima in training of neural networks", which shows that on finite datasets one can construct datasets or initialization schemes that trap networks in bad local minima.

  5. 2017

    Sobolev training for neural networks

    In 2017, he coauthors "Sobolev Training for Neural Networks" with Czarnecki, Simon Osindero, Max Jaderberg and Pascanu, a method that trains networks on a target function's derivatives along with its values. The paper tests it on regression, on distilling policies of an agent playing Atari and on synthetic gradients.

  6. 2022–2025

    AlphaTensor and polytope discovery

    He is one of the authors of the October 2022 Nature paper on AlphaTensor, an AlphaZero-based agent that discovers provably correct matrix multiplication algorithms and beats the best known complexity for many matrix sizes. In January 2025, he and nine coauthors present an AI system that generates complex polytopes, matching or surpassing the best known bounds for the Hirsch conjecture, the k-neighbourly problem and the longest monotone paths problem.

  7. Since September 2024

    Quant research director at Qube

    In September 2024, after more than eight years at Google DeepMind, where his roles include staff research scientist in the deep learning group, he joins the London quant hedge fund Qube Research & Technologies as Quant Research Director.

  1. Google had the most people shaping AI every year from 2014 to 2025Our estimates of which organizations employed the most people shaping AI from 1961 to 2025, with a year-by-year chart.
  2. The United States employs an estimated 53% of people in our data with a known employment countryOur estimates of where AI careers are concentrated now and how countries’ shares of dated positions changed over time.
  3. 46% of living people in our data with known birth and employment countries work outside their birth countryOur estimate of cross-border careers in AI, from country of birth and the country of each current organization.
  4. More people shaping AI studied at Massachusetts Institute of Technology than at any other institutionOur estimates of where the people shaping AI studied and where their careers led, from education records in cited biographies.

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 Grzegorz Swirszcz yet. Send us a free-licensed photo.

  1. Grzegorz Świrszcz – Warsaw.AI warsaw.ai. Checked 19 October 2026.
  2. Grzegorz Świrszcz - The Mathematics Genealogy Project mathgenealogy.org. Checked 19 October 2026.
  3. Former Google DeepMind AI expert joins London quant Qube | Hedgeweek hedgeweek.com. Checked 19 October 2026.
  4. Winning the KDD Cup Orange Challenge with Ensemble Selection proceedings.mlr.press. Checked 19 October 2026.
  5. [1611.06310] Local minima in training of neural networks arxiv.org. Checked 19 October 2026.
  6. [1706.04859] Sobolev Training for Neural Networks arxiv.org. Checked 19 October 2026.
  7. Discovering novel algorithms with AlphaTensor — Google DeepMind deepmind.google. Checked 19 October 2026.
  8. Discovering faster matrix multiplication algorithms with reinforcement learning | Nature nature.com. Checked 19 October 2026.
  9. [2502.05199] Advancing Geometry with AI: Multi-agent Generation of Polytopes arxiv.org. Checked 19 October 2026.