CV
Education
National University of Singapore (NUS), Ph.D. in Mechanical Engineering, Jan 2021 - Present
- Advisor: Prof. Marcelo H. Ang Jr.
- GPA: 4.33 / 5.00
- Relevant Coursework: Machine Vision, Deep Learning for Robotics, Digital Human
Huazhong University of Science and Technology (HUST), B.E. in Mechanical Engineering, Sep 2014 - Jun 2018
- GPA: 3.80 / 4.00
Experience
Xiaomi Technology (Embodied Intelligence Team) Research Intern May 2025 - Present
- Mentor: Dr. Long Chen
- Core contributor to the MiMo-Embodied Foundation Model project. Led data curation and evaluation for the Embodied AI domain, including Chain-of-Thought (CoT) and RL data.
- Achieved SOTA performance across 17 benchmarks, validating scaled data and training strategies.
- Leveraging Xiaomi AI Glass as a low-cost, scalable platform to collect diverse egocentric human manipulation videos for Vision-Language-Action (VLA) model training.
- Exploring World Models and Latent Action Models to transfer human dexterity to generalist robots.
Microsoft Research Asia (MSRA) Research Intern May 2024 - Present
- Mentors: Dr. Jiaolong Yang & Dr. Yu Deng
- Core author of the VITRA project; developed the first fully-automated human 3D VLA data generation pipeline transforming real-life human videos into robot-aligned training data.
- Engineered a high-fidelity 3D Motion Labeling pipeline integrating HaWoR for hand reconstruction and a modified MegaSAM for metric-scale camera tracking.
- Processed 26 million frames to curate 1 million atomic VLA episodes with dense 3D supervision.
- Engineered a Transformer-based RL model for dexterous robotic grasping, achieving a 91.2% success rate on seen objects in simulation.
Skills
- Research: Embodied AI, Vision-Language-Action (VLA) Models, 3D Reconstruction, Visual SLAM, Reinforcement Learning
- Programming: Python, C++, PyTorch, CUDA
- Tools: Isaac Gym, ROS
- Languages: English (Professional), Chinese (Native)