About Me

Hi 👋 I am Xin Yin (殷鑫).

I am a fifth-year Ph.D. student at Zhejiang University, supervised by Prof. Chao Ni and Prof. Xiaohu Yang. My research interest includes Large Language Model, Coding Agent, and Post-Train.

I am currently a research intern on the RL Team at Moonshot AI (Kimi), where I lead efforts on WebDev for the Kimi K3-series Model:

• Around Code Arena | WebDev (3D, Game, Website), I conduct task decomposition, failure attribution, and capability-gap analysis to guide data construction;
• I build verifiable RL environments, designing rewards based on execution and visual interaction to turn task completion into stable, reproducible training signals for long-horizon tasks;
• I am responsible for Post-Train (Large-Scale RL + SFT), including training strategy design and optimization, improving the long-horizon coding ability of Kimi’s flagship model in these scenarios.

🔥 News

• 2026.09:  🎉 Kimi K2.8 Preview is coming! Technical Report.
• 2026.09:  🎉 Step 5 Preview is coming! Technical Report.
• 2026.05:  🎉 Step 3.7 Flash is coming! Technical Report.
• 2025.10:  🎉 I am awarded National Scholarship!
• 2025.06:  🎉 I am funded by Zhejiang University’s Program for Cultivating Outstanding Doctoral Dissertations!

💼 Experience

• Moonshot AI (Kimi)
  Research Intern · RL Team · WebDev (3D, Game, Website)
  • Kimi K2.8 Preview: Performance close to K3, with more efficient thinking
• StepFun
  Research Intern · Post-Training Team · Terminal Agent
  • Step 5 Preview: Advancing the Pareto Frontier
  • Step 3.7 Flash: A high-efficiency Flash model for real-world agents
• Tongyi Lab
  Research Intern · ModelScope Team · Agent Framework & LLM Post-training
  • MS-Agent : core developer, leading Code Genesis and post-training for long-horizon coding agents
  • Ultron : led self-evolving multi-agent memory, skills, and a shared Harness
  • Hello-Agents : core contributor on agent self-evolution

📝 Publications

# denotes co-first author or first student author
Representative papers: 23 CCF-A papers, 3 TH-CPL-A papers, 2 JCR-Q1 papers

Selected Publications

• RepoDistill: Distilling Repository Knowledge through Compression-Aware Budget Allocation and Policy Optimization.
  Xin Yin, Zixiang Ding, Yiang Zhang, Qiang Wang, Rui Wang, Chao Ni, Zhe Cui.
  In Findings of the 64th Annual Meeting of the Association for Computational Linguistics (ACL’26 Findings). (CCF-A)
• Aligning with Human Coding Preferences for Improving Code Generation.
  Xin Yin, Chao Ni, Xiaohu Yang.
  In Proceedings of the 34th ACM International Conference on the Foundations of Software Engineering (FSE’26). (CCF-A)
• Improving the Ability of Pre-trained Language Model by Imparting Large Language Model’s Experience.
  Xin Yin, Chao Ni, Xinrui Li, Xiaohu Yang.
  In Journal of Systems and Software (JSS’25). (JCR-Q1)
• Enhancing LLM’s Ability to Generate More Repository-Aware Unit Tests Through Precise Context Injection.
  Xin Yin, Chao Ni, Xinrui Li, Liushan Chen, Guojun Ma, Xiaohu Yang.
  In Proceedings of the 40th IEEE/ACM Automated Software Engineering Conference (ASE’25). (CCF-A)
• What You See Is What You Get: Attention-based Self-guided Automatic Unit Test Generation.
  Xin Yin, Chao Ni, Xiaodan Xu, Xiaohu Yang.
  In Proceedings of the 47th IEEE/ACM International Conference on Software Engineering (ICSE’25). (CCF-A)
• Multitask-based Evaluation of Open-Source LLM on Software Vulnerability.
  Xin Yin, Chao Ni, Shaohua Wang.
  In IEEE Transactions on Software Engineering (TSE’24). (CCF-A)
• ThinkRepair: Self-Directed Automated Program Repair.
  Xin Yin, Chao Ni, Shaohua Wang, Zhenhao Li, Limin Zeng, Xiaohu Yang.
  In Proceedings of the 33rd ACM SIGSOFT International Symposium on Software Testing and Analysis (ISSTA’24). (CCF-A)
• Learning-based Models for Vulnerability Detection: An Extensive Study.
  Chao Ni, Xin Yin#, Liyu Shen, Shaohua Wang.
  In Empirical Software Engineering (EMSE’25). (JCR-Q1)
• Abundant Modalities Offer More Nutrients: Multi-Modal-Based Function-level Vulnerability Detection.
  Chao Ni, Xin Yin#, Xinrui Li, Xiaodan Xu, Zhi Yu.
  In ACM Transactions on Software Engineering and Methodology (TOSEM’25). (CCF-A)
• Distinguishing Look-Alike Innocent and Vulnerable Code by Subtle Semantic Representation Learning and Explanation.
  Chao Ni, Xin Yin#, Kaiwen Yang, Dehai Zhao, Zhenchang Xing, Xin Xia.
  In Proceedings of the 31st ACM International Conference on the Foundations of Software Engineering (FSE’23). (CCF-A)
• Adaptive Mutation Scheduling with Deep Reinforcement Learning for Smart Contract Fuzzing.
  Qianqian Pang, Xin Yin#, Tingting Bi, Lingfeng Bao, Chao Ni, Xiaohu Yang.
  In Proceedings of the 34th ACM International Conference on the Foundations of Software Engineering (FSE’26). (CCF-A)
• RepoGenesis: Benchmarking End-to-End Microservice Generation from Readme to Repository.
  Zhiyuan Peng, Xin Yin#, Pu Zhao, Fangkai Yang, Lu Wang, Ran Jia, Xu Chen, Saravan Rajmohan, Dongmei Zhang.
  In Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (ACL’26 Main). (CCF-A)
• SolEval: Benchmarking Large Language Models for Repository-level Solidity Smart Contract Generation.
  Zhiyuan Peng, Xin Yin#, Rui Qian, Peiqin Lin, YongKang Liu, Hao Zhang, Chenhao Ying, Yuan Luo.
  In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing (EMNLP’25 Main). (TH-CPL-A)
• PrefGen: A Preference-Driven Methodology for Secure Yet Gas-Efficient Smart Contract Generation.
  Zhiyuan Peng, Xin Yin#, Zijie Zhou, Chenhao Ying, Chao Ni, Yuan Luo.
  In Proceedings of the 40th IEEE/ACM Automated Software Engineering Conference (ASE’25). (CCF-A)

Selected Collaborative Publications

• Breaking Waiting: Accelerating Android GUI Testing via Widget Readiness Analysis.
  Xinrui Li, Xin Yin, Chao Ni, Xiaoyu Sun, Jue Wang, Xiaohu Yang.
  In ACM Transactions on Software Engineering and Methodology (TOSEM’26). (CCF-A)
• Hunk-Constrained DPO: Segment-Level Optimization for Secure and Correct LLM Code Generation.
  Qianshuo Huang, Xin Yin, Xinrui Li, Chao Ni.
  In ACM Transactions on Software Engineering and Methodology (TOSEM’26). (CCF-A)
• UGround: Towards Unified Visual Grounding with Unrolled Transformers.
  Rui Qian, Xin Yin, Chuanhang Deng, Zhiyuan Peng, Jian Xiong, Wei Zhai, Dejing Dou.
  In Proceedings of the 43rd International Conference on Machine Learning (ICML’26). (CCF-A)
• JUnitGenie: A Framework for Path-Sensitive Unit Test Generation with Large Language Models.
  Dianshu Liao, Xin Yin, Shidong Pan, Chao Ni, Zhenchang Xing, Xiaoyu Sun.
  In 48th IEEE/ACM International Conference on Software Engineering (ICSE’26 Demonstrations Track). (CCF-A)
• Navigating the Labyrinth: Path-Sensitive Unit Test Generation with Large Language Models.
  Dianshu Liao, Xin Yin, Shidong Pan, Chao Ni, Zhenchang Xing, Xiaoyu Sun.
  In Proceedings of the 40th IEEE/ACM Automated Software Engineering Conference (ASE’25). (CCF-A)
• Reasoning to Attend: Try to Understand How <SEG> Token Works.
  Rui Qian, Xin Yin, Dejing Dou.
  In Proceedings of the 2025 IEEE/CVF Computer Vision and Pattern Recognition Conference (CVPR’25). (CCF-A)
• MOCA: A Hierarchical Semantic-Enhanced Code Edit Framework for Multilingual Code Co-Evolution.
  Zhihao Gong, Zeyu Sun, Xin Yin, Yizhou Chen, Qingyuan Liang, Guoqing Wang, Jie Zhang, Dan Hao.
  In ACM Transactions on Software Engineering and Methodology (TOSEM’26). (CCF-A)
• SepPrune: Structured Pruning for Efficient Deep Speech Separation.
  Yuqi Li, Kai Li, Xin Yin, Zhifei Yang, Junhao Dong, Zeyu Dong, Chuanguang Yang, Yingli Tian, Yao Lu.
  In Proceedings of the 40th Annual AAAI Conference on Artificial Intelligence (AAAI’26). (CCF-A)
• PlayCoder: Making LLM-Generated GUI Code Playable.
  Zhiyuan Peng, Wei Tao, Xin Yin, Chenhao Ying, Yuan Luo, Yiwen Guo.
  In Proceedings of the 34th ACM International Conference on the Foundations of Software Engineering (FSE’26). (CCF-A)
• Input Reduction Enhanced LLM-based Program Repair.
  Boyang Yang, Luyao Ren, Xin Yin, Jiadong Ren, Haoye Tian, Shunfu Jin.
  In Proceedings of the 48th IEEE/ACM International Conference on Software Engineering (ICSE’26). (CCF-A)
• FigmaBench: Evaluating Design-to-Code Generation in Real-World Handoff Scenarios.
  Ziyang Wang, Ziyang Liu, Xin Yin, Chao Zhang, Zhe Cui, Yue Lu.
  In Findings of the 2026 Conference on Empirical Methods in Natural Language Processing (EMNLP’26 Findings). (TH-CPL-A)
• Tree-CoT-RT: An Explainable Multi-Path Tree-Guided Chain-of-Thought and Reinforcement Learning Framework for Aspect Sentiment Quad Prediction.
  Hao Zhang, Jiahao Wang, Zhenke Duan, Xin Yin, Haichuan Hu, Hualong Chen, SUYI, Congqing He, Yike Tan, Yu-N Cheah.
  In Findings of the 64th Annual Meeting of the Association for Computational Linguistics (ACL’26 Findings). (CCF-A)
• ArkEval: Benchmarking and Evaluating Automated Code Repair for ArkTS.
  Bang Xie, Senjian Zhang, Zhiyuan Peng, Wei Chen, Xin Yin, Chenhao Ying, Yuan Luo.
  In Proceedings of the 41st IEEE/ACM Automated Software Engineering Conference (ASE’26). (CCF-A)
• Pre-training CLIP against Data Poisoning with Optimal Transport-based Matching and Alignment.
  Tong Zhang, Kuofeng Gao, Jiawang Bai, Leo Yu Zhang, Xin Yin, Zonghui Wang, Shouling Ji, Wenzhi Chen.
  In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing (EMNLP’25 Main). (TH-CPL-A)

🎖 Honors and Awards

• 2025.10, 浙江大学国家奖学金 (National Scholarship)
• 2025.06, 浙江大学争创优秀博士学位论文资助

💬 Academic Services

• Journal Reviewer: IEEE Transactions on Software Engineering (TSE), ACM Transactions on Software Engineering and Methodology (TOSEM), Empirical Software Engineering (EMSE), Automated Software Engineering (ASE)
• Conference Reviewer: ICSE 2026 (Shadow PC), AAAI 2026 (PC), ICLR 2026, NeurIPS 2026, AAAI 2027 (PC), ICLR 2027