Research Summary

My research primarily focuses on Code LLMs and Coding Agents, including their training, evaluation, and inference-time optimization. More recently, I have been working on recursive self-improvement, particularly on enabling models to train themselves and optimize their own agent harnesses. I have published 30+ papers in these areas, including an ICSE 2025 paper that received the ACM SIGSOFT Distinguished Paper Award.

Education

  • Ph.D. in Computer Science and Engineering, August 2023 - July 2027 (expected)
    • The Chinese University of Hong Kong, Hong Kong, China
    • Supervisor: Prof. Michael R. Lyu
  • M.E. in Computer Science and Technology, September 2021 - June 2023
    • Harbin Institute of Technology, Shenzhen, China
    • Supervisor: Prof. Cuiyun Gao
  • B.E. in Cyberspace Security, September 2017 - June 2021
    • Harbin Institute of Technology, Weihai, China
    • GPA: 91.98/100, Rank: 1/34

Experience

  • December 2025 - Present: Research Intern
    • ByteDance Douyin AI4SE
    • Mentor: Dr. Shilin He
    • Working on coding-agent SFT and RL, including long-horizon agents with compaction and subagents, agentic RL training and credit assignment, and automated post-training
  • March 2021 - August 2021: Research Intern
    • Huawei Noah’s Ark Lab, Shenzhen, China
    • Mentor: Dr. Pengyun Wang
    • Worked on long-tailed image classification

Selected Publications

  • [EMNLP 2026] SWE-MeM: Learning Adaptive Memory Management for Long-Horizon Coding Agents
    • Shuzheng Gao, Wenhao Zeng, Zhaojian Yu, Jianqiao Wangni, Chaozheng Wang, Kai Cai, Shilin He and Michael R. Lyu
    • Oral Presentation
  • [ICSE 2026] SEER: Enhancing Chain-of-Thought Code Generation through Self-Exploring Deep Reasoning
    • Shuzheng Gao, Chaozheng Wang, Cuiyun Gao and Michael R. Lyu
  • [LLM4Code 2026] ContextPilot: Code Context Engineering with Memory-Augmented Exploration Agents
    • Shuzheng Gao, Chaozheng Wang, Shuqing Li, Yun Peng and Michael R. Lyu
  • [ICSE 2025] Search-Based LLMs for Code Optimization
    • Shuzheng Gao, Cuiyun Gao, Wenchao Gu and Michael R. Lyu
    • ACM SIGSOFT Distinguished Paper Award
  • [ICSE 2024] Learning in the Wild: Towards Leveraging Unlabeled Data for Effectively Tuning Pre-trained Code Models
    • Shuzheng Gao, Wenxin Mao, Cuiyun Gao, Li Li, Xing Hu, Xin Xia and Michael R. Lyu
  • [ASE 2023] What Makes Good In-context Demonstrations for Code Intelligence Tasks with LLMs?
    • Shuzheng Gao, Xin-Cheng Wen, Cuiyun Gao, Wenxuan Wang, Hongyu Zhang and Michael R. Lyu
  • See the complete list on the Publications page.

Teaching

Courses I served as a teaching assistant:

  • COMP 2005 Digital Logic Design
    • Fall 2021, Harbin Institute of Technology, Shenzhen
  • ENGG 1100 Problem Solving By Programming
    • Fall 2023 and Fall 2024, The Chinese University of Hong Kong
  • CSCI 3100 Software Engineering
    • Spring 2024 and Spring 2025, The Chinese University of Hong Kong

Academic Service

  • Program Committee: FORGE 2025, Internetware 2025
  • Reviewer: TSE, TOSEM, KAIS, EMSE, ASEJ, Neural Networks
  • Sub-reviewer:
    • Journals: TSE, TOSEM, TASLP
    • Conferences: ICSE 2027, ICSE 2026, ACL 2026, ISSRE 2025, ACL 2025, FSE 2025, ISSTA 2024, FSE 2024, ASE 2023, Internetware 2023, ICSE-NIER 2023, ICSME 2022, SANER 2022, ASE-Tool 2022, IJCAI 2022

Honors and Awards

  • SAC Highlight Award, EMNLP, 2026
  • ACM SIGSOFT Distinguished Paper Award, ICSE, 2025
  • Internetware Distinguished Reviewer Award, 2025
  • Distinguished Paper Award, ASE Industry Challenge, 2023
  • Binxing Fang Scholarship, 2023
  • Best Student Paper Award, ACAIT, 2022
  • Outstanding Graduate of Harbin Institute of Technology, 2021
  • National Scholarship, Ministry of Education, 2019

Skills

  • Languages: Mandarin (native), English (fluent; IELTS 7.0)