Skip to main content

Work & study

Work & experience

A few chapters, most recent first.

  1. September 2025 - presentCurrent

    Member of Technical Staff

    Microsoft AI, Post-training

    Working on MAI-Code models, with a particular focus on reinforcement learning with verifiable rewards for mathematics and the sciences. I use Lean to build scalable training datasets grounded in machine-checked proofs, turning formal verification into a reliable reward signal.

  2. September 2024 - 2025

    Science Lead

    Azure's AI Safety Team, Responsible AI

    Led the science pillar within Azure's AI Safety Team. Trained and shipped production-scale models powering the integrated safety systems in Azure OpenAI and Azure AI Foundry, supporting responsible AI protections at a scale of billions of requests per day.

  3. 2021 - 2024

    Research Scientist

    Microsoft GenAI

    Led model evaluation for the development of Phi-3, Microsoft's family of small language models.

  4. 2018 - 2021

    Applied Scientist

    Microsoft, ExP

    Worked as a scientist on Microsoft's experimentation platform team. My main contribution was a compute-agnostic implementation of the delta method for variance estimates with multiple aggregation levels.

  5. 2013 - 2018

    Mathematics PhD

    Cornell University

    Got my PhD in the stable representation theory of certain categories under the supervision of Martin Kassabov.