Yezeyuan Zheng is a research assistant at the HKQAI Lab, focusing on material structure prediction, material generation, and the development of an LLM-powered agent system. He is exploring material generation and crystalline structure prediction using reinforcement learning within a custom environment, as well as employing diffusion models and language models for material generation. His research goal is to apply machine learning to tackle complex multidisciplinary problems, many of which involve significant uncertainty. To address this uncertainty, Yezeyuan builds simulated environments and utilizes advanced deep learning models to approximate the Q-function using reinforcement learning. This approach aims to generate a set of trajectories representing the best possible actions in the context of material science.