Derek Murray

Software Engineer at Google DeepMind

Redwood City, California, United States

About

I develop novel distributed systems for large-scale data processing. In this field, I focus on two areas: increasing the expressiveness and performance over existing systems. Currently, I work on data infrastructure at Lacework, which involves processing huge volumes of customer data to produce insights about cloud security. Before that, I was a core member of the TensorFlow team, in Google Brain. At Microsoft Research, I developed Naiad, which is the first system to support incremental, iterative, and interactive computations on a distributed cluster. Previously, I received a PhD from the University of Cambridge for my work on the CIEL distributed execution engine. For CIEL, I developed a new programming language (Skywriting) and a runtime that could execute imperative and functional code in a distributed system. Skywriting makes it simple to express iteration and recursion in a distributed workflow, and CIEL executes these programs reliably and efficiently. During my PhD I also dabbled in programming languages (the Steno compiler for declarative queries) and operating system virtualization (using the Xen hypervisor).

Experience

  • Principal Software Engineer at Google DeepMind
    Nov 2025 - Present · 8 mos

    Various kinds of agents

  • Senior Staff Software Engineer at Google
    Sep 2023 - Nov 2025 · 2 yrs 3 mos

  • Software Engineer at Lacework
    Jul 2021 - Aug 2023 · 2 yrs 2 mos

  • Principal Architect at Microsoft
    May 2020 - Jul 2021 · 1 yr 3 mos

    Architect on the AI Frameworks team within Microsoft Azure, focused on training support in the ONNX Runtime.

  • Senior Staff Software Engineer at Google
    Dec 2014 - May 2020 · 5 yrs 6 mos

    Helped to create TensorFlow: Google's machine learning system for deep neural networks and more.