About me

Hi, thanks for visiting my homepage!

I'm a postdoctoral researcher working with Jeff Schneider at the Robotics Institute (RI), Carnegie Mellon University. My recent focus is on Tokamak plasma control in nuclear fusion. Our goal is to design better feedback control policy via RL and broader ML-based approaches.

I received my PhD degree in Computer Science and Engineering at The Chinese University of Hong Kong (CUHK), where I’m blessed to be under co-supervision of Farzan Farnia and Ho-fung Leung. My PhD thesis focuses on provably efficient reinforcement learning (RL) and its applications to the evaluation and selection of generative AI models. I obtained a Bachelor’s degree in Statistics from Sun Yat-sen University (SYSU) in Guangzhou, China.

Research Interests

  • Web development icon

    Tokamak Plasma Control

    How can we design better feedback control of Tokamak plasma via RL?

  • Web development icon

    Provably efficient RL

    How optimal behaviors be sample-efficiently learned?

  • Web development icon

    Generative AI

    How can RL and online learning assist the training, generation, and evaluation of generative models?

Bio

My full CV can be downloaded here.

Professional Appointments

  1. Carnegie Mellon University

    2025 — Present

    - Postdoctoral Reseacher

    - PI: Jeff Schneider

Education

  1. The Chinese University of Hong Kong (CUHK)

    2021 — 2025

    - Doctor of Philosophy in Computer Science and Engineering

    - Supervisors: Farzan Farnia, Ho-fung Leung

  2. Sun Yat-sen University (SYSU)

    2017 — 2021

    - Bachelor of Sciences in Statistics

    - Supervisor: Li Xia

Research & Visiting Experience

  1. Guest PhD Student @ DeLTA Lab, DIKU, University of Copenhagen

    Oct 2023 - Jan 2024 (Host: Professor Mohammad Sadegh Talebi)

  2. Visiting PhD Student @ Huawei Noah's Ark Lab

    May — August 2022
  3. Algorithm Engineer @ Timi Studio Group, Tencent

    April — June 2021

Invited Talks

  1. An Information Theoretic Approach to Interaction-Grounded Learning [slides]

    @ Scool Team, Inria, Jan 11, 2024

  2. Provably (More) Efficient Offline RL with Options [slides]

    @ DeLTA Seminar, Oct 12, 2023

Service

  1. Reviewer

    NeurIPS (2024-), ICLR (2025-), ICML (2025), AISTATS (2025-)

Publications

Contact