PeopleofSuperintelligence

Builder

Andy Davis

Andy Davis is a builder of AI systems whose work includes TensorFlow and tools for distributed machine learning. He coauthored a sparsely gated mixture-of-experts model, research on dynamic control flow, and the MLIR compiler infrastructure. In 2025, he coauthored EQuARX, a quantized AllReduce method for distributed machine-learning acceleration. As of October 2026, he works at Google Research on machine intelligence and networking.

Works
7
Citations
21,948
h-index
6

Publication numbers from OpenAlex, as of 15 October 2026.

Timeline

  1. 2015–2016

    Building TensorFlow

    He coauthors the 2015 TensorFlow interface paper and the 2016 OSDI paper on its large-scale machine-learning system. The API and reference implementation are released as open-source software in November 2015.

  2. 2017

    A mixture-of-experts layer

    In 2017, he coauthors a sparsely gated mixture-of-experts architecture with a trainable gate and models that reach 137 billion parameters.

  3. 2018

    TensorFlow dynamic control flow

    In 2018, he coauthors a TensorFlow programming model that supports dynamic control flow in distributed machine learning.

  4. 2021

    MLIR compiler infrastructure

    In 2021, he coauthors the MLIR paper on reusable compiler infrastructure for domain-specific compilers and heterogeneous hardware.

  5. 2025

    EQuARX for distributed learning

    In 2025, he coauthors EQuARX, a quantized AllReduce implementation within the XLA compiler for TPUs that targets distributed machine-learning acceleration.

  6. As of October 2026

    Research at Google

    As of October 2026, he works at Google Research on machine intelligence and networking.

Sources and credits

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

  1. TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems research.google. Checked 15 October 2026.
  2. TensorFlow: A System for Large-Scale Machine Learning usenix.org. Checked 15 October 2026.
  3. Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer research.google. Checked 15 October 2026.
  4. Dynamic Control Flow in Large-Scale Machine Learning arxiv.org. Checked 15 October 2026.
  5. FAQ mlir.llvm.org. Checked 15 October 2026.
  6. MLIR: Scaling Compiler Infrastructure for Domain Specific Computation research.google. Checked 15 October 2026.
  7. EQuARX: Efficient Quantized AllReduce in XLA for Distributed Machine Learning Acceleration research.google. Checked 15 October 2026.
  8. EQuARX: Efficient Quantized AllReduce in XLA for Distributed Machine Learning Acceleration arxiv.org. Checked 15 October 2026.
  9. Andy Davis research.google. Checked 15 October 2026.