PeopleofSuperintelligence

Researcher

Nilesh Tripuraneni

Nilesh Tripuraneni is a machine learning researcher at Google DeepMind who works on language models, including Gemini. A Harvard physics graduate, he completed an M.Phil. at the University of Cambridge under Zoubin Ghahramani, and his paper Lost Relatives of the Gumbel Trick won an ICML 2017 honorable mention. His PhD at the University of California, Berkeley, advised by Michael Jordan, ended with a 2022 dissertation on transfer learning and robust learning, after a 2020 NeurIPS paper on the theory of transfer learning. He appears among the authors of the Gemini and Gemini 1.5 reports in 2023 and 2024 and of the Michelangelo long-context evaluation in 2024, and in July 2026 he coauthored a paper on estimators for randomized control trials.

Badges

Timeline

  1. 2017

    Cambridge and Magnetic HMC

    A physics graduate of Harvard University, he completes an M.Phil. at the University of Cambridge under the supervision of Zoubin Ghahramani. In 2017, he coauthors Magnetic Hamiltonian Monte Carlo with Mark Rowland, Ghahramani and Richard Turner, a generalization of Hamiltonian Monte Carlo that uses non-canonical Hamiltonian dynamics.

  2. 2017

    An ICML 2017 honorable mention

    Lost Relatives of the Gumbel Trick, written with Matej Balog, Zoubin Ghahramani and Adrian Weller, earns an honorable mention among the ICML 2017 awards. The paper derives a family of methods, of which the Gumbel trick is one member, for sampling from a discrete distribution or estimating its partition function.

  3. 2020

    Theory of transfer learning

    In 2020, he publishes On the Theory of Transfer Learning: The Importance of Task Diversity at NeurIPS with Michael I. Jordan and Chi Jin. The paper gives statistical guarantees for transfer through a feature representation shared across tasks, built on a new notion of task diversity.

  4. 2022

    A Berkeley PhD under Jordan

    He completes a PhD in computer science at the University of California, Berkeley in 2022, advised by Michael Jordan. His dissertation, Learning Beyond the Standard Model (of Data), develops theory for transfer learning and robust learning, asking why pretraining is so effective and what the limits of learning are for extremely heavy-tailed distributions.

  5. 2023–2024

    The Gemini and Gemini 1.5 reports

    He appears among the authors of the Gemini report in December 2023, which introduces the multimodal Gemini family of Ultra, Pro and Nano models. He is listed again on the Gemini 1.5 report in March 2024, which covers models that recall and reason over millions of tokens of context.

  6. September 2024

    Michelangelo long-context evaluation

    In September 2024, he coauthors Michelangelo with Kiran Vodrahalli, Santiago Ontanon and colleagues. The paper introduces a synthetic long-context reasoning evaluation for large language models, built on a framework called Latent Structure Queries.

  7. July 2026

    Optimal estimators for randomized trials

    In July 2026, he coauthors Towards Optimal Estimators for Randomized Control Trials with Harsh Parikh, Dhruv Madeka, Michael I. Jordan and others. The paper proposes a framework for identifying optimal estimators within families of randomized controlled trials.

  8. As of October 2026

    Researcher at Google DeepMind

    He is a researcher at Google DeepMind, where he works on language models, including Gemini, and on broader topics in machine learning and artificial intelligence.

Career

  1. NowGoogle DeepMindResearcher

Education

  1. 2022University of California, BerkeleyPhD
  2. Harvard University
  3. University of CambridgeM.Phil.

From the biography.

  1. 16% of AI career moves in the 2020s went from a university to a companyIncludes Nilesh Tripuraneni. 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. 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.
  3. 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.
  4. 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.

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 Nilesh Tripuraneni yet. Send us a free-licensed photo.

  1. Nilesh Tripuraneni nileshtrip.github.io. Checked 22 October 2026.
  2. Magnetic Hamiltonian Monte Carlo proceedings.mlr.press. Checked 22 October 2026.
  3. ICML 2017 Awards icml.cc. Checked 22 October 2026.
  4. Lost Relatives of the Gumbel Trick arxiv.org. Checked 22 October 2026.
  5. On the Theory of Transfer Learning: The Importance of Task Diversity arxiv.org. Checked 22 October 2026.
  6. Learning Beyond the Standard Model (of Data) www2.eecs.berkeley.edu. Checked 22 October 2026.
  7. Gemini: A Family of Highly Capable Multimodal Models arxiv.org. Checked 22 October 2026.
  8. Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context arxiv.org. Checked 22 October 2026.
  9. Michelangelo: Long Context Evaluations Beyond Haystacks via Latent Structure Queries arxiv.org. Checked 22 October 2026.
  10. Towards Optimal Estimators for Randomized Control Trials arxiv.org. Checked 22 October 2026.
  11. Nilesh Tripuraneni – MLG Cambridge mlg.eng.cam.ac.uk. Checked 22 October 2026.