2026
AI researcher / Beijing
Yi Jing
Building agents that produce knowledge alongside people.
I'm a fourth-year undergraduate at Tsinghua University and an incoming PhD student at Peking University's Institute of Computational Linguistics, advised by Prof. Zhifang Sui.
Currently, I'm a research intern at Tencent Hy (Project UP), working on post-training and LLM agents.
I want to build agents that take part in producing knowledge, from professional work to scientific research, and keep learning from practice. More broadly, I hope AI and people can grow together: systems that make people sharper thinkers, not ones that think for them.
Selected research
All publications2026
2025
Latest writing
All writingBringing Decision Models into Knowledge Work
Can professional judgment scale by becoming cheaper rather than slower and larger? Using legal and financial benchmarks, this note examines how low-cost decision models expand coverage and route difficult cases to long reasoning.
How Agents Learn from Environments and Feedback
Understanding the agent training loop through Reward, State, and Credit: Reward sets the direction of learning, State sets where training happens, and Credit decides which actions the outcome is attributed to.
Background
Xinya College & Computer Science, Tsinghua University.
Previously, I worked on interpretability and reinforcement learning at THUKEG, human-centered AI at Maryland's CLIP Lab, and AI for neuroscience at THBI and Oxford's FMRIB.
Collaborators & mentors
Prof. Juanzi Li (THUKEG), Prof. Jordan Boyd-Graber (CLIP), Prof. Qiyuan Tian (THBI), and Prof. Wenchuan Wu (FMRIB).