Background

Experience & education

RL post-training, adaptive computation, and agents that learn from experience.

Research experience

  1. May – Aug 2026

    Microsoft

    Research Intern · Frontier Tuning AI team

    Invited to join the team to extend research on test-time inference and hallucination control. Developed ForkPilot, a self-evolving search policy that learns from completed trajectories to improve long-horizon agent efficiency.

  2. Aug 2024 – Present

    Virginia Tech

    Graduate Research Assistant

    Developed predictors and detectors for LLM selection and hallucination risk, alongside RL post-training and adaptive test-time computation.

Education

  1. 2024 – Present

    Virginia Tech

    Ph.D. in Computer Science

    Advisor: Prof. Dawei Zhou

  2. 2022 – 2024

    University of North Carolina at Chapel Hill

    M.S. in Biostatistics

    Advisor: Prof. Yun Li

  3. 2018 – 2022

    Zhejiang University

    B.S. in Statistics

    Advisor: Prof. Peng Zhang