PeopleofSuperintelligence

Researcher

Peter Battaglia

Also known as Peter W. Battaglia

Peter Battaglia is a Senior Director of Research at Google DeepMind whose graph-network models taught machines to simulate physics and then to forecast weather. After a psychology PhD at the University of Minnesota and research at MIT's Department of Brain and Cognitive Sciences, he was first author of the 2016 Interaction Networks paper and the 2018 paper 'Relational inductive biases, deep learning, and graph networks'. His 2021 paper 'Learning Mesh-Based Simulation with Graph Networks', written with Tobias Pfaff, Meire Fortunato and Alvaro Sanchez-Gonzalez, was an ICLR 2021 outstanding paper. He was on the Google DeepMind GraphCast team that won the 2024 MacRobert Award. He leads initiatives at the intersection of AI and environmental sustainability, and his team developed the WeatherNext family of weather models.

Timeline

  1. By 2013

    Intuitive physics at MIT

    He earns a PhD in psychology at the University of Minnesota and later works as a postdoc and research scientist in MIT's Department of Brain and Cognitive Sciences. In 2013, he coauthors 'Simulation as an engine of physical scene understanding' with Jessica Hamrick and Joshua Tenenbaum, which proposes an 'intuitive physics engine' that makes predictions by approximate, probabilistic simulation.

  2. 2016

    Interaction Networks at NIPS

    In 2016, he is first author of 'Interaction Networks for Learning about Objects, Relations and Physics', written with Razvan Pascanu, Matthew Lai, Danilo Rezende and Koray Kavukcuoglu and listed among DeepMind's NIPS papers. The model reasons over graph-structured data and serves as a general-purpose learnable physics engine that simulates n-body, bouncing ball and string systems over thousands of time steps.

  3. 2018

    Graph networks for relational reasoning

    In 2018, he is first author of 'Relational inductive biases, deep learning, and graph networks'. It introduces the graph network as a building block for the AI toolkit with a strong relational inductive bias, and comes with an open-source software library for building graph networks.

  4. 2020

    Learning to simulate physics

    In 2020, he coauthors 'Learning to Simulate Complex Physics with Graph Networks' with Alvaro Sanchez-Gonzalez, Jonathan Godwin, Tobias Pfaff, Rex Ying and Jure Leskovec. Its Graph Network-based Simulators represent a physical system as particles in a graph and compute dynamics through learned message-passing, across fluids, rigid solids and deformable materials.

  5. 2021

    An ICLR 2021 outstanding paper

    In 2021, he coauthors 'Learning Mesh-Based Simulation with Graph Networks' with Tobias Pfaff, Meire Fortunato and Alvaro Sanchez-Gonzalez, an ICLR 2021 outstanding paper. Its MeshGraphNets framework predicts the dynamics of physical systems such as aerodynamics, structural mechanics and cloth, and runs 1-2 orders of magnitude faster than the simulation it is trained on.

  6. 2022–2023

    GraphCast weather forecasting

    A GraphCast preprint with him among its authors appears in December 2022; the model predicts hundreds of weather variables over 10 days at 0.25 degree resolution in under one minute. When the study is published in November 2023, he is quoted as Google DeepMind's research director, and GraphCast is shown to be more precise and faster than the High-Resolution Forecast (HRES), the industry gold standard.

  7. July 2024

    The MacRobert Award

    In July 2024, GraphCast, developed by Google DeepMind, is announced as the 2024 MacRobert Award winner of the Royal Academy of Engineering. He is a research scientist on the winning team, whose model generates in 45 seconds a forecast that takes traditional methods over an hour on a supercomputer.

  8. As of October 2026

    Senior Director at DeepMind

    He is Senior Director of Research at Google DeepMind, where he leads initiatives at the intersection of AI and environmental sustainability, and his team developed the WeatherNext family of AI weather models. In October 2025, WeatherNext identifies about a week in advance that a small tropical disturbance will intensify into a Category 5 hurricane bound for Jamaica.

Career

  1. NowGoogle DeepMindResearch Director
  2. Massachusetts Institute of TechnologyPostdoc

Education

  1. University of MinnesotaPhD

From the biography.

  1. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other universityIncludes Peter Battaglia. 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.
  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 Peter Battaglia yet. Send us a free-licensed photo.

  1. Peter Battaglia - Structured models of physics, objects, and scenes neurips.cc. Checked 22 October 2026.
  2. Simulation as an engine of physical scene understanding pmc.ncbi.nlm.nih.gov. Checked 22 October 2026.
  3. Interaction Networks for Learning about Objects, Relations and Physics arxiv.org. Checked 22 October 2026.
  4. DeepMind Papers @ NIPS (Part 1) deepmind.google. Checked 22 October 2026.
  5. Relational inductive biases, deep learning, and graph networks arxiv.org. Checked 22 October 2026.
  6. Learning to Simulate Complex Physics with Graph Networks arxiv.org. Checked 22 October 2026.
  7. Learning mesh-based simulations sites.google.com. Checked 22 October 2026.
  8. ICLR Poster Learning Mesh-Based Simulation with Graph Networks iclr.cc. Checked 22 October 2026.
  9. GraphCast: Learning skillful medium-range global weather forecasting arxiv.org. Checked 22 October 2026.
  10. DeepMind says its new AI system is the world’s most accurate 10-day weather forecaster thenextweb.com. Checked 22 October 2026.
  11. 2024 MacRobert Award finalists macrobertaward.raeng.org.uk. Checked 22 October 2026.
  12. MacRobert Award for Engineering 2024 – The Worshipful Company of Engineers engineerscompany.org.uk. Checked 22 October 2026.
  13. EPS Colloquium - Peter Battaglia, Google DeepMind - Harvard EPS eps.harvard.edu. Checked 22 October 2026.
  14. At the Simons Science Summit, Leaders Explore How AI Is Transforming Scientific Discovery – Simons Foundation simonsfoundation.org. Checked 22 October 2026.