I am a third-year PhD candidate of Intelligent Processor Research Center(智能处理器研究中心), Institute of Computing Technology, supervised by Rui Zhang, Yuanbo Wen and Xishan Zhang.

Previously, as an undergraduate student at Harbin Institute of Technology, I completed a research internship at SCIR, working on Dialogue System, advised by Wei-Nan Zhang.

My research focuses on Recursive Self-Improvement (RSI) through AI4Chip and AI4Sys: AI improves chip design and system software, enabling more capable AI.

  • Reinforcement Learning for Model Post-Training
  • Self-Verifying and Self-Improving Agents
  • Agent Systems for Reliably Scaling Frontier Discovery
RSI Recursive Self-Improvement

AI uses self-evolving agents and reinforcement learning to improve chips and system software. Advances across this hardware and software stack enable stronger AI, which drives the next iteration.

AI
Self-Evolving Agents Reinforcement Learning
System Software
Chip Design
Stronger AI

Open to collaboration, academic exchange, and internship opportunities. kcxain@gmail.com

Contributions

News

  • [2026.04] QiMeng-PRepair has been accepted to ACL 2026.
  • [2025.11] Awarded the Excellent Prize of the President Scholarship(所长优秀奖) at ICT, CAS.
  • [2025.09] QiMeng-MuPa has been accepted to NeurIPS 2025.

Publications

Education

2024.08 - Present
Institute of Computing Technology
Master
2020.09 - 2024.06
Harbin Institute of Technology
Undergraduate

Experience

2026.01 - Present
Cambricon
AI Research Intern

Conferences Attended

Patents

  • 一种自验证式的探索预训练语言模型中明喻知识的方法及其系统

    张伟男,马龙轩,柯昌鑫,刘挺

Competitions

Honors

  • [2026.10] Huang Kun–Cambricon Qingjin Award(黄昆-寒武纪青衿奖), 20th National Semiconductor and Integrated Technology Conference
  • [2026.05] Outstanding Model Student(三好学生标兵), UCAS
  • [2025.11] Excellent Prize of the President Scholarship(所长优秀奖), ICT, CAS

Services

  • Reviewer: EMNLP 2023, NeurIPS 2026, AAAI 2027, ICLR 2027

Visitors