Yijia Fan

I am Yijia Fan, a fourth-year undergraduate in Computer Science and Technology at Sun Yat-sen University, supervised by Prof. Keze Wang. I will join MMLab at The Chinese University of Hong Kong as a Ph.D. student in Fall 2027, advised by Prof. Xiangyu Yue.

I am a research intern at Microsoft Research Asia – Visual Computing Group, working on video generation, unified multimodal models, agent skills, and recursive self-improvement (self-evolving agents). I was a research intern with the SenseNova team at SenseTime, where I worked on native reflection in unified models with reinforcement learning (UMM-Reflection). I am currently also interning at an embodied AI startup, building embodied agent systems. I have previously interned at the Human Cyber-Physical Intelligence Lab at Sun Yat-sen University.

My research interests include unified multimodal models, visual language models, reinforcement learning, self-evolving agentic systems, and embodied agent systems. I am particularly interested in aligning visual perception with complex reasoning and exploring unified generation toward a world model that integrates perception, reasoning, and creation.

Contact: fanyj28@mail2.sysu.edu.cn