A Little About Me

Hi, I’m Wenjie Shu. I’m currently a researcher at 2077 AI, a nonprofit organization.

I received my B.E. in Information Engineering from the University of Electronic Science and Technology of China (UESTC) in 2025, where I worked closely with Prof. Liangjian Deng. I also spent time at the Hong Kong University of Science and Technology (HKUST) as a visiting student and research assistant, working with Prof. Harry Yang and Prof. Qifeng Chen. I’ve been fortunate to collaborate with Dr. Xiaogang Xu and Dr. Rui-Jie Zhu, and my collaborators also include Tim G. J. Rudner, Xianghong Fang, and Dehan Kong.

My research centers on image and video generation, visual reasoning benchmarks, and agents with recursive self-improvement. I’m interested in how generative models understand and solve visual problems, how we can evaluate those abilities, and how models and agents can improve through interaction and feedback.

I’d like to use this blog to share research notes, observations from experiments, and questions I’m still thinking about. Some posts may be works in progress; writing them down is a way to make the ideas clearer and open up discussion.

I’m currently applying to PhD programs and am always open to research collaborations. If you’re working on related problems, or would like to discuss something I’ve posted, feel free to reach out at wenjieshu2003@gmail.com.