PeopleofSuperintelligence

Researcher

Borja Ibarz

Also known as Borja Ibarz Gabardos

Borja Ibarz is a Google DeepMind researcher whose work connects reinforcement learning, algorithmic reasoning and AI for science. He co-authored research on human-feedback rewards, neural algorithm learners and normalizing flows for free-energy estimates. In 2024, he co-authored CLRS-Text and contributed to AlphaProof and a FunSearch update on competitive programming; AlphaProof and AlphaGeometry 2 reached silver-medal level at the IMO. As of October 2026, he is affiliated with Google DeepMind as a researcher in formal mathematical reasoning.

Works
23
Citations
862
h-index
12
Badges

Publication numbers from OpenAlex, as of 19 October 2026.

Timeline

  1. 2018

    Learning rewards from people

    In 2018, he co-authors research that learns reward functions from human preferences and demonstrations, then trains reinforcement-learning agents on nine Atari games.

  2. 2021

    Spatial embeddings for agents

    A 2021 DeepMind patent names him as an inventor on a neural network that turns an agent’s movement through an environment into spatial embeddings.

  3. 2022

    A generalist algorithm learner

    His 2022 paper introduces a graph neural network processor that executes a wide range of algorithms, from sorting and searching to dynamic programming and geometry.

  4. September 2023

    Free energy with normalizing flows

    In September 2023, he co-authors a normalizing-flow model that estimates Gibbs free energies by sampling the isobaric-isothermal ensemble, including tests on cubic and hexagonal ice.

  5. June 2024

    Text benchmarks for reasoning

    In June 2024, he co-authors CLRS-Text, a text version of algorithmic traces that generates examples for thirty tasks and evaluates language models as algorithm executors.

  6. July 2024

    AlphaProof and the IMO

    At the IMO in July 2024, he contributes to AlphaProof, a reinforcement-learning system for formal mathematical reasoning. Paired with AlphaGeometry 2, the system solves four of six problems and reaches silver-medal standard.

  7. December 2024

    Competitive programming with FunSearch

    In December 2024, he contributes to the FunSearch update on combinatorial competitive programming. The report applies the method to amplify human performance in competitive programming.

  8. As of October 2026

    Google DeepMind researcher

    As of October 2026, he is affiliated with Google DeepMind as an AI researcher. A March 2026 Nature paper lists him as an author on AlphaProof research and describes the system’s reinforcement-learning method for formal proofs.

Career

  1. Since 2020Google DeepMindAI researcher
  2. 2004–2008Universidad Rey Juan Carlos

From the biography and OpenAlex affiliations (the years of papers that list the organization), so some years are approximate.

Most-cited works

  1. Map-based models in neuronal dynamics2010 · Physics Reports · 328 citations
  2. Reward learning from human preferences and demonstrations in Atari2018 · 117 citationsWith Geoffrey Irving, Jan Leike and Shane Legg
  3. Synchronization and propagation of bursts in networks of coupled map neurons2006 · Chaos An Interdisciplinary Journal of Nonlinear Science · 80 citations
  4. Bursting regimes in map-based neuron models coupled through fast threshold modulation2008 · Physical Review E · 57 citations
  5. Growing Critical: Self-Organized Criticality in a Developing Neural System2018 · Physical Review Letters · 54 citations

From OpenAlex, as of 19 October 2026.

  1. The median h-index is 54 at universities and 19 at frontier AI labsIncludes Borja Ibarz. The median h-index is 54 at universities and 19 at frontier AI labs. We compare publishing careers, citations and moves to estimate how universities and frontier labs contribute to AI research.
  2. 14% of AI career moves in the 2020s went from a university to a companyIncludes Borja Ibarz. Our estimates of how university-to-company career transitions changed by decade, including roles added alongside an earlier job.
  3. Google has the highest combined h-index among AI organizationsIncludes Google. Our estimates of combined and median h-index at AI organizations, using current affiliations and OpenAlex records.
  4. Google alumni founded or led 81 other organizations, the most among the labs we compareIncludes Google. We trace 194 organizations founded or led by AI lab alumni, and 202 career moves between the labs.

Colleagues

People with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Borja Ibarz yet. Send us a free-licensed photo.

  1. Reward learning from human preferences and demonstrations in Atari arxiv.org. Checked 18 October 2026.
  2. WO2021228985A1 - Generating spatial embeddings by integrating agent motion and optimizing a predictive objective - Google Patents patents.google.com. Checked 18 October 2026.
  3. A Generalist Neural Algorithmic Learner proceedings.mlr.press. Checked 18 October 2026.
  4. Estimating Gibbs free energies via isobaric-isothermal flows — Google DeepMind deepmind.google. Checked 18 October 2026.
  5. The CLRS-Text Algorithmic Reasoning Language Benchmark arxiv.org. Checked 18 October 2026.
  6. AI achieves silver-medal standard solving International Mathematical Olympiad problems — Google DeepMind deepmind.google. Checked 18 October 2026.
  7. FunSearch: Making new discoveries in mathematical sciences using Large Language Models — Google DeepMind deepmind.google. Checked 18 October 2026.
  8. Olympiad-level formal mathematical reasoning with reinforcement learning | Nature nature.com. Checked 18 October 2026.