PeopleofSuperintelligence

Researcher

Gintare Karolina Dziugaite

Gintare Karolina Dziugaite is a machine-learning researcher whose work links generative networks, learning theory and machine unlearning. She studied mathematics at Warwick and Cambridge, earned a machine-learning PhD at Cambridge in 2019, and took part in research programs at the Simons Institute and the Institute for Advanced Study. Her stochastic convex optimization paper shared an ICML Best Paper award in 2024; later papers examined teacher-network size as a measure of data complexity and machine unlearning with public data. As of October 2026, she is a senior research scientist at Google DeepMind, an adjunct professor at McGill, and an associate industry member of Mila.

Badges

Timeline

  1. 2015

    Generative neural networks with MMD

    In 2015, she coauthors “Training generative neural networks via Maximum Mean Discrepancy optimization” with Daniel M. Roy and Zoubin Ghahramani. The paper trains a generator by minimizing an unbiased estimate of maximum mean discrepancy.

  2. 2017

    A Simons Institute program

    In winter 2017, she joins the Simons Institute’s Foundations of Machine Learning program as a long-term participant.

  3. 2019

    A Cambridge PhD and fellowship

    She completes a PhD in machine learning at Cambridge in 2019 under Zoubin Ghahramani, after studying mathematics at Warwick and Cambridge. In summer 2019, she returns to Simons as a fellow in its Foundations of Deep Learning program.

  4. 2020

    An IAS membership

    In 2020, she is a member of the Institute for Advanced Study in Princeton and participates in its Special Year on Optimization, Statistics, and Theoretical Machine Learning.

  5. 2024

    An ICML Best Paper

    In 2024, she coauthors “Information Complexity of Stochastic Convex Optimization: Applications to Generalization, Memorization, and Tracing” with Idan Attias, Mahdi Haghifam, Roi Livni and Daniel M. Roy. ICML gives it a Best Paper award; the paper studies how memorization and learning interact in stochastic convex optimization.

  6. 2025

    Teacher networks and data complexity

    In 2025, she and Daniel M. Roy introduce a PAC-Bayes notion of data complexity based on the smallest teacher network that achieves a given excess risk.

  7. 2026

    Unlearning with public data

    In 2026, she coauthors “Unlearning with Asymmetric Sources: Improved Unlearning-Utility Trade-off with Public Data,” which introduces Asymmetric Langevin Unlearning to use public data to reduce the privacy cost of unlearning.

  8. As of October 2026

    Google DeepMind and Mila

    As of October 2026, she is a senior research scientist at Google DeepMind in Toronto, an adjunct professor at McGill’s School of Computer Science, and an associate industry member of Mila. Before joining Google, she leads the Trustworthy AI program at Element AI (ServiceNow).

Career

  1. NowGoogle DeepMindSenior Research Scientist
  2. NowMcGill UniversityAdjunct Professor

Education

  1. 2019University of CambridgePhD in Machine Learning
  2. University of Warwick

From the biography.

  1. 16% of AI career moves in the 2020s went from a university to a companyIncludes Gintare Karolina Dziugaite. 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.
  2. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other universityIncludes Gintare Karolina Dziugaite. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other university. We compare doctorate and 5+ year employment links, career timing and current workplaces for 187 major-prize and 1,111 best-paper winners.
  3. 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.
  4. 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.

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 Gintare Karolina Dziugaite yet. Send us a free-licensed photo.

  1. Gintare Karolina Dziugaite – MLG Cambridge mlg.eng.cam.ac.uk. Checked 21 October 2026.
  2. Gintare Karolina Dziugaite gkdz.org. Checked 21 October 2026.
  3. ICML 2024 Awards icml.cc. Checked 21 October 2026.
  4. Information Complexity of Stochastic Convex Optimization: Applications to Generalization, Memorization, and Tracing proceedings.mlr.press. Checked 21 October 2026.
  5. The Size of Teachers as a Measure of Data Complexity: PAC-Bayes Excess Risk Bounds and Scaling Laws proceedings.mlr.press. Checked 21 October 2026.
  6. Unlearning with Asymmetric Sources: Improved Unlearning-Utility Trade-off with Public Data proceedings.mlr.press. Checked 21 October 2026.
  7. Gintare Karolina Dziugaite | Mila mila.quebec. Checked 21 October 2026.