I am interested in creating robots and agents that are capable by training, steerable by inference, and improved by experience. Today, a robot’s capabilities are no longer fixed at training time: they can be steered at inference time (diffusion-model guidance, CAR), generalised to new environments (skill learning, SaMI), and further improved through interaction (fast online adaptation, ARLPE).
If you’d like to chat about any of this, feel free to email me!
|
Postdoctoral Fellow @ National University of Singapore (NUS) - (2025-present)
Postdoctoral Fellow at the Smart Systems Institute, supervised by Prof. Harold Soh. Research on Embodied AI safety and inference-time alignment for diffusion models.
|
|
Research Intern @ Beijing Institute for General Artificial Intelligence (BIGAI) - (2025)
Research on learning robot manipulation from offline expert datasets (human demonstrations and cross-embodiment data).
|
|
University of Edinburgh - (2023-2024)
Project: Skill-aware Mutual Information Optimisation for Generalisation in Reinforcement Learning
Visiting student in the Autonomous Agents Research Group, supervised by Prof. Stefano V. Albrecht; funded by the World's Top Universities Strategic Cooperation Fellowship (in Chinese, “世界顶尖大学战略合作计划”).
|
|
Harbin Institute of Technology - (2019-2025)
Thesis: Causal Reinforcement Learning Generalisation for Healthcare Agents
Completed my PhD at the Faculty of Computing, HIT; defended my viva in March 2025.
GPA: 92.53/100
|
|
Harbin Engineering University - (2015-2019)
B.Eng. in Internet of Things Engineering, College of Computer Science and Technology; awarded Outstanding Graduate and Outstanding Graduation Thesis (2019).
GPA: 88.95/100.
|