Robotics Developer & Researcher,
building intelligent systems from embedded hardware to learned world models.

I'm a Master's student in Technology Innovation (Robotics Track) at the University of Washington (GPA: 3.83/4.0), with a B.S. in Software Engineering from Tongji University, and I completed exchange coursework at UC Berkeley. My work spans robotics, embedded systems, and full-stack software development. My research interests include robot learning, reinforcement and imitation learning, world models, safety verification, human-robot interaction, brain-machine interfaces and neural decoding.

Education

University of WashingtonSep 2025 to Mar 2027

M.S. in Technology Innovation (Robotics Track), GPA 3.83/4

Courses: Managing Data and Signal Processing, Robotic Sensing and Mobility, Neuroscience (NEUSCI 401)

Tongji UniversitySep 2021 to Jun 2025

B.S. in Software Engineering, GPA 88.25/100

Courses: Linear Algebra, Probability and Mathematical Statistics, Statistics Analysis and Modeling, Data Structures, Computer Graphics

Experience

Timeplus, ShanghaiJul 2024 to Feb 2025

Backend Development Intern

  • Developed the C++ core engine of Proton, a real-time streaming database for high-throughput, low-latency data processing, extending the streaming type system with 6 new data types and ensuring compatibility across 48 streaming aggregation functions.
  • Resolved 3 long-standing bugs in distributed streaming storage, including merge-operation failures and data consistency issues between real-time insertion and query results.

Skills

Languages: C++, Python  |  ML: PyTorch, scikit-learn, world models, RL and imitation learning  |  Robotics: ROS2, LeRobot, Isaac Lab, NVIDIA Jetson  |  Tools: Docker, Git

Also: Java, C#, JavaScript, CUDA, MoveIt, Fusion 360

Publications & Manuscripts

Do Joint Audio-Video Generation Models Understand Physics?

The paper introduces AV-Phys Bench, a benchmark for physical commonsense in joint audio-video generation.

Co-author. NeurIPS 2026. Paper

TACT: Tracking Asking Cost over Time for Personalized Clarification in Human-Robot Interaction

Chenming Ge, Fan Zhang

Manuscript in preparation. Code

WM-Verdict: Benchmarking the Reliability of World-Model Verdicts about Policies

Manuscript in preparation.

Algorithm Structure for Real-time General Game Playing Agents

Independent research, 2022. Paper