PeopleofSuperintelligence

Researcher

H. Brendan McMahan

Also known as Brendan McMahan, Hugh Brendan McMahan

Brendan McMahan is a Google researcher whose 2016 paper named federated learning, a way to train a shared model from updates computed on mobile devices. He earned a PhD at Carnegie Mellon University in 2006 and led the author list of a 2013 KDD paper on ad click-through-rate prediction. In 2017, he and Daniel Ramage announced a test of federated learning in Gboard, and he coauthored a paper on differentially private recurrent language models. A 2019 survey he coauthored with Peter Kairouz and others set out open problems in the field, and in 2022 he and Abhradeep Thakurta announced a Gboard model trained with federated learning and a formal differential privacy guarantee. He is a principal research scientist at Google, where he leads efforts on decentralized and privacy-preserving machine learning.

Now at
Google
Badges

Timeline

  1. 2006

    A PhD at Carnegie Mellon

    He earns a PhD in computer science from Carnegie Mellon University in 2006, advised by Avrim Blum and Geoffrey J. Gordon. His dissertation is titled "Robust planning in domains with stochastic outcomes, adversaries, and partial observability".

  2. 2013

    Ad click prediction at KDD

    In 2013, he leads the author list of "Ad Click Prediction: a View from the Trenches", a paper for the 19th ACM SIGKDD conference. It presents case studies from a deployed click-through-rate prediction system, including an FTRL-Proximal online learning algorithm and per-coordinate learning rates.

  3. 2016

    Naming federated learning

    In 2016, he leads the author list of "Communication-Efficient Learning of Deep Networks from Decentralized Data", with Eider Moore, Daniel Ramage, Seth Hampson and Blaise Agüera y Arcas. The paper terms the decentralized approach Federated Learning and presents a method based on iterative model averaging, which cuts the required communication rounds by 10-100x compared with synchronized stochastic gradient descent.

  4. 2017

    Federated learning in Gboard

    In April 2017, he and Daniel Ramage introduce federated learning on the Google Research blog as a way for mobile phones to collaboratively learn a shared prediction model while the training data stays on the device. Google is testing the approach in Gboard on Android, the Google Keyboard, to improve query suggestions.

  5. 2017

    Private recurrent language models

    In 2017, he coauthors "Learning Differentially Private Recurrent Language Models" with Daniel Ramage, Kunal Talwar and Li Zhang. The paper adds user-level privacy protection to the federated averaging algorithm and shows that large recurrent language models can be trained with user-level differential privacy at a negligible cost in predictive accuracy.

  6. 2019–2020

    A survey of open problems

    In December 2019, he coauthors "Advances and Open Problems in Federated Learning" with Peter Kairouz and dozens of other authors, a survey of recent advances and open problems. In 2020, he gives an ICML workshop keynote on the paper, which 22 Google researchers and 36 academics from 24 institutions wrote together.

  7. 2022

    DP-FTRL in Gboard

    In February 2022, he and Abhradeep Thakurta announce on the Google Research blog a Spanish-language next-word-prediction model for Gboard, trained with federated learning and the DP-FTRL algorithm under a formal differential privacy guarantee. Google describes it as, to its knowledge, the first production neural network trained directly on user data announced with a formal DP guarantee.

  8. As of October 2026

    Principal research scientist at Google

    He is a principal research scientist at Google, where he leads efforts on decentralized and privacy-preserving machine learning. In 2024, he coauthors "Federated Learning in Practice: Reflections and Projections" with Katharine Daly, Hubert Eichner, Peter Kairouz, Daniel Ramage and Zheng Xu, and in February 2026 he gives the Richard M. Karp Distinguished Lecture at the Simons Institute.

Career

  1. NowGooglePrincipal research scientist

Education

  1. 2006Carnegie Mellon UniversityPhD

From the biography.

  1. 16% of AI career moves in the 2020s went from a university to a companyIncludes H. Brendan McMahan. 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 H. Brendan McMahan yet. Send us a free-licensed photo.

  1. Hugh McMahan - The Mathematics Genealogy Project mathgenealogy.org. Checked 21 October 2026.
  2. Responsibly Improving AI with Privacy-Sensitive Data: Principles, Theory, and Practice | Richard M. Karp Distinguished Lecture simons.berkeley.edu. Checked 21 October 2026.
  3. Ad Click Prediction: a View from the Trenches research.google. Checked 21 October 2026.
  4. Communication-Efficient Learning of Deep Networks from Decentralized Data arxiv.org. Checked 21 October 2026.
  5. Federated Learning: Collaborative Machine Learning without Centralized Training Data research.google. Checked 21 October 2026.
  6. Learning Differentially Private Recurrent Language Models arxiv.org. Checked 21 October 2026.
  7. Advances and Open Problems in Federated Learning arxiv.org. Checked 21 October 2026.
  8. ICML Keynote Session 5: Advances and Open Problems in Federated Learning, by Brendan McMahan (Google) icml.cc. Checked 21 October 2026.
  9. Federated Learning with Formal Differential Privacy Guarantees research.google. Checked 21 October 2026.
  10. Federated Learning in Practice: Reflections and Projections arxiv.org. Checked 21 October 2026.