Yiming Xiong

I am Yiming Xiong (熊一鸣), a Data Science Master's student at the University of British Columbia. My research interest lies in self-improving agents for autonomous scientific research: how AI systems can autonomously search for novel improvements, design experiments to validate them, and accumulate reusable knowledge to improve future problem-solving and research processes.

I am fortunate to work with Shengran Hu and Jeff Clune on exploring the autonomous discovery of agent memory designs for continual learning. I am currently an intern at the

Kimi post-training team, where I work on model behavior analysis, RL pipeline design, and data construction to improve base models' knowledge accumulation and self-improvement capabilities. I am also a research associate at

MIT, collaborating with Ao Qu and Paul Liang on orchestrating multi-agent AI scientists for reinforcement learning research.

Selected Research Program

Learning to Continually Learn via Meta-learning Agentic Memory Designs

Yiming Xiong, Shengran Hu, Jeff Clune

🏆 Best Paper @ MemAgents Workshop · Outstanding Paper @ RSI Workshop, ICLR 2026

Publications

Yiming Xiong, Shengran Hu, Jeff Clune. Learning to Continually Learn via Meta-learning Agentic Memory Designs. arXiv:2602.07755, 2026. arXiv

Kimi Team (including Yiming Xiong). Kimi K3: Open Frontier Intelligence. Technical Report, 2026. arXiv

Youliang Yuan, Wenxuan Wang, Qingshuo Guo, Yiming Xiong, Chihao Shen, Pinjia He. Does ChatGPT Know That It Does Not Know? Evaluating the Black-Box Calibration of ChatGPT. LREC-COLING 2024, pp. 5191–5201. ACL Anthology