PeopleofSuperintelligence

Researcher

Matej Balog

Matej Balog is a staff research scientist at Google DeepMind, where they build AI agents that discover algorithms. They studied mathematics and computer science at Oxford before beginning a machine-learning PhD in 2015 in the Cambridge-Tübingen program. In 2016 they were first author of DeepCoder, a deep-learning system that learns to write programs from input-output examples. At DeepMind they contributed to AlphaTensor, FunSearch, and AlphaEvolve. In 2017 they shared an ICML Best Paper honorable mention for Lost Relatives of the Gumbel Trick.

Works
18
Citations
1,282
h-index
7

Publication numbers from OpenAlex, as of 15 October 2026.

Timeline

  1. 2011–2015

    An Oxford thesis on Mondrian processes

    From 2011 they study mathematics and computer science at Oxford; their final thesis, done with Yee Whye Teh, explores new applications of the Mondrian process.

  2. 2015–2020

    PhD in Cambridge and Tübingen

    In 2015 they begin a machine-learning PhD in the Cambridge-Tübingen program, supervised by Zoubin Ghahramani.

  3. 2016

    DeepCoder learns from examples

    DeepCoder, a deep-learning system that learns to write programs from input-output examples, is first posted to arXiv in 2016 with them as first author.

  4. 2017

    An ICML 2017 honorable mention

    At ICML 2017, Lost Relatives of the Gumbel Trick, written by Matej Balog with Nilesh Tripuraneni, Zoubin Ghahramani and Adrian Weller, receives a Best Paper honorable mention.

  5. 2022

    Matrix multiplication with AlphaTensor

    AlphaTensor, DeepMind's system for discovering matrix multiplication algorithms, is introduced in October 2022 with Balog among its contributors.

  6. 2023

    A FunSearch paper in Nature

    Nature publishes the FunSearch paper in December 2023; it uses large language models to search for mathematical discoveries, and they co-author it.

  7. Since 2024

    Staff research scientist at DeepMind

    They work as a staff research scientist at Google DeepMind, building AI agents that discover algorithms and helping develop AlphaEvolve.

Sources and credits

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

  1. Matej Balog | alphaXiv alphaxiv.org. Checked 15 October 2026.
  2. [1611.01989] DeepCoder: Learning to Write Programs arxiv.org. Checked 15 October 2026.
  3. ICML 2017 Awards icml.cc. Checked 15 October 2026.
  4. Prize winning papers from MLG at ICML 2017 – MLG Cambridge mlg.eng.cam.ac.uk. Checked 15 October 2026.
  5. Discovering novel algorithms with AlphaTensor — Google DeepMind deepmind.google. Checked 15 October 2026.
  6. Discovering faster matrix multiplication algorithms with reinforcement learning | Nature nature.com. Checked 15 October 2026.
  7. Mathematical discoveries from program search with large language models | Nature nature.com. Checked 15 October 2026.
  8. AlphaEvolve: A Gemini-powered coding agent for designing advanced algorithms — Google DeepMind deepmind.google. Checked 15 October 2026.