ABOUT · RESEARCH

Post-training foundation models and agents.

I am Jingsheng Gao (高景盛), a researcher working on large language models, post-training, and multimodal learning. My work connects data, learning algorithms, and evaluation to make foundation models more capable, reliable, and useful in real-world settings.

My current interests include LLM post-training and reinforcement learning, retrieval-augmented generation, data generation, dialogue intelligence, multimodal learning, and parameter-efficient adaptation. My doctoral research at Shanghai Jiao Tong University has been conducted at the School of Electronic Information and Electrical Engineering, focusing on foundation models, dialogue systems, and multimodal learning. Before my doctoral studies, I earned my B.S. in Physics from SJTU’s School of Physics and Astronomy.

I have previously collaborated with Suncheng Xiang, Mingye Xie, Jiacheng Ruan, Ke Ji, Baoyuan Wang, Yixin Lian, and Bin Dai on research spanning language models, dialogue intelligence, efficient adaptation, and multimodal learning.

I welcome research collaborations around LLM post-training, reinforcement learning, RAG, dialogue systems, and multimodal learning. In my free time, I enjoy swimming, board games, and photography.