Yaxin Du

PhD Student

Shanghai Jiao Tong University

I am a fourth-year direct-entry Ph.D. student at Shanghai Jiao Tong University, advised by Prof. Siheng Chen. I am currently exploring job opportunities and would be happy to connect. Outside research, I enjoy meeting new people, food, travel, and discovering new things.

Research Overview

My recent work centers on how language-model-based agents can operate reliably in realistic environments, improve through training, and be evaluated on open-ended tasks.

Tool Use and Environment Interaction. I study how agents interact with tools and external environments, with recent work on MCP-based tool learning, web browsing, and environment-grounded agent benchmarks.
Training Agents and Coding Models. I work on training strategies for agentic and coding-oriented language models, including role-wise multi-agent training, coder-verifier co-evolution, and mid-training for coding LLMs.
Evaluation in Open-Ended Tasks. I build benchmarks and data-centric resources for studying how agents and coding models behave on realistic, feedback-driven tasks.

News

Jun 2026 Released HyperTool: Beyond Step-Wise Tool Calls for Tool-Augmented Agents.
Jun 2026 Released MIRA: Mid-training Rubric Anchoring for Source-Aware Data Selection.
May 2026 Released DataMaster: Data-Centric Autonomous AI Research.
May 2026 Three papers were accepted to ICML 2026: G2-Reader, MCP-Persona, and NTK-Selector.
Apr 2026 Two papers accepted to ACL 2026: MCP-Flow and VLMGuard-R1.
Apr 2026 Released the InCoder paper series. [Paper]
Feb 2026 Released G²-Reader.
Jan 2026 Joined iQuest Lab, Ubiquant, as a Research Intern, working on foundation model training and coding LLM mid-training.
Jan 2026 InfoMosaicBench was accepted to ICLR 2026.
Dec 2025 Presented SWE-Dev at the NeurIPS 2025 Workshop.
Sep 2025 Joined TikTok AI Innovation Center as a Research Intern, mentored by Qian Liu.

Publications

Selected Publications

arXiv 2026 HyperTool: Beyond Step-Wise Tool Calls for Tool-Augmented Agents
Yaxin Du*, Yifan Zhou*, Yujie Ge, Jiajun Wang, Xianghe Pang, Shuo Tang, Tuney Zheng, Bryan Dai, Jian Yang, Siheng Chen
arXiv preprint arXiv:2606.13663 [Paper]
arXiv 2026 DataMaster: Data-Centric Autonomous AI Research
Yaxin Du*, Xiyuan Yang*, Zhifan Zhou, Wanxu Liu, Zixing Lei, Zimeng Chen, Fenyi Liu, Haotian Wu, Yuzhu Cai, Zexi Liu, Xinyu Zhu, WenHao Wang, Linfeng Zhang, Chen Qian, Siheng Chen [Paper]
ICLR 2026 InfoMosaic-Bench: Evaluating Multi-Source Information Seeking in Tool-Augmented Agents
Yaxin Du, Yuanshuo Zhang, Xiyuan Yang, Yifan Zhou, Cheng Wang, Gongyi Zou, Xianghe Pang, Wenhao Wang, et al.
arXiv preprint arXiv:2510.02271 [Paper]
ICML 2026 G2-Reader: Dual Evolving Graphs for Multimodal Document QA
Yaxin Du*, Junru Song*, Yifan Zhou*, Cheng Wang, Jiahao Gu, Zimeng Chen, Menglan Chen, Wen Yao, Yang Yang, Ying Wen, Siheng Chen [Paper]
arXiv 2025 SWE-Dev: Evaluating and Training Autonomous Feature-Driven Software Development
Yaxin Du*, Yuzhu Cai*, Yifan Zhou, Cheng Wang, Yu Qian, Xianghe Pang, Qian Liu, Yue Hu, Siheng Chen
NeurIPS 2025 DL4C Workshop [Paper] [Code]
ACL 2025 Findings FedDQC: Data Quality Control in Federated Instruction-tuning of Large Language Models
Yaxin Du, Rui Ye, Fengting Yuchi, Wanru Zhao, Jingjing Qu, Yanfeng Wang, Siheng Chen [Paper]
ICLR 2024 Workshop Enhancing Data Quality in Federated Fine-tuning of Large Language Models
Wanru Zhao*, Yaxin Du*, Nicholas Donald Lane, Siheng Chen, Yanfeng Wang [Paper]
ICML 2026 NTK-Selector: Selecting Auxiliary Data via Neural Tangent Kernels for Low-Resource Domains
Pingjie Wang, Hongcheng Liu, Yusheng Liao, Ziqing Fan, Yaxin Du, Shuo Tang, Yanfeng Wang, Yu Wang [Paper]
ICLR 2025 Self-evolving Multi-agent Collaboration Networks for Software Development
Yue Hu, Yuzhu Cai, Yaxin Du, Xinyu Zhu, Xiangrui Liu, Zijie Yu, Yuchen Hou, Shuo Tang, Siheng Chen
arXiv 2025 MAS-GPT: Training LLMs to Build LLM-based Multi-Agent Systems
Rui Ye, Shuo Tang, Rui Ge, Yaxin Du, Zhenfei Yin, Siheng Chen, Jing Shao [Paper]
KDD 2024 OpenFedLLM: Training Large Language Models on Decentralized Private Data via Federated Learning
Rui Ye, Wenhao Wang, Jingyi Chai, Dihan Li, Zexi Li, Yinda Xu, Yaxin Du, Yanfeng Wang, Siheng Chen [Paper]
NeurIPS 2024 FedLLM-Bench: Realistic Benchmarks for Federated Learning of Large Language Models
Rui Ye, Rui Ge, Xinyu Zhu, Jingyi Chai, Yaxin Du, Yang Liu, Yanfeng Wang, Siheng Chen [Paper]
ICLR 2024 Fake It Till Make It: Federated Learning with Consensus-Oriented Generation
Rui Ye, Yaxin Du, Zhenyang Ni, Siheng Chen, Yanfeng Wang [Paper]

Other Publications

arXiv 2026 MIRA: Mid-training Rubric Anchoring for Source-Aware Data Selection
H. Wang, Y. Du, J. Yang, J. Wu, S. Liu, Y. Zhang, P. Wang, S. Chen, T. Zheng, M. Zhou, X. Liu [Paper]
arXiv 2026 EvoMaster: A Foundational Agent Framework for Building Evolving Autonomous Scientific Agents at Scale
X. Zhu, Y. Cai, Z. Liu, C. Wang, F. Li, W. Jin, W. Liu, Z. Bing, B. Zheng, J. Chai, S. Tang, R. Ye, Y. Du, X. Pang, Y. Du, T. Miao, Y. Zhang, R. Liao, Z. Ding, L. Zhang, Y. Wang, W. E, S. Chen [Paper]
arXiv 2025 VLMGuard-R1: Proactive Safety Alignment for VLMs via Reasoning-Driven Prompt Optimization
M. Chen, X. Pang, J. Dong, W. H. Wang, Y. Du, S. Chen [Paper]
arXiv 2025 BrowseMaster: Towards Scalable Web Browsing via Tool-Augmented Programmatic Agent Pair
X. Pang, S. Tang, R. Ye, Y. Du, Y. Du, S. Chen [Paper]
ACL 2026 MCP-Flow: Facilitating LLM Agents to Master Real-World, Diverse and Scaling MCP Tools
W. Wang, P. Niu, Z. Xu, Z. Chen, J. Du, Y. Du, X. Pang, K. Huang, Y. Wang, et al. [Paper]
arXiv 2025 Federated Instruction Tuning of LLMs with Domain Coverage Augmentation
Z. Wang, Y. Du, Z. Qian, S. Chen
IET RSN 2024 Radar-based Human Activity Recognition Using Denoising Techniques to Enhance Classification Accuracy
R. Yu, Y. Du, J. Li, A. Napolitano, J. Le Kernec
arXiv 2026 InCoder-32B: Code Foundation Model for Industrial Scenarios
J. Yang, W. Zhang, J. Wu, J. Cheng, S. Guo, H. Wang, W. Gu, Y. Du, J. Li, F. Xu, et al. [Paper]
arXiv 2026 InCoder-32B-Thinking: Industrial Code World Model for Thinking
J. Yang, W. Zhang, J. Wu, J. Cheng, T. Zheng, F. Xu, W. Gu, L. Jing, Y. Du, J. Li, et al. [Paper]
RadarConf 2022 A ViT Approach for Short-range Behaviour Recognition Using Radar Signals
Y. Du, B. Li, J. Li, F. Fioranelli, J. Le Kernec
CIE Radar 2021 Radar-based Human Activity Classification with Cyclostationarity
Y. Du, J. Li, Z. Li, R. Yu, A. Napolitano, F. Fioranelli, J. Le Kernec

Research Experience

Jan 2026 - Present Research Intern, iQuest Lab, Ubiquant
Working on foundation model training, with a recent focus on mid-training for coding LLMs.
Sep 2025 - Jan 2026 Research Intern, TikTok AI Innovation Center
Mentored by Qian Liu; worked on LLM agents, coding-related language model research, and related systems problems.

Blog

May 2026

Training Multi-Agent Systems: From MAS Generators to Co-Evolving Agents

This post summarizes how I think about training multi-agent systems, including MAS generation, component-level training, and co-evolution. I also discuss why training MAS is fundamentally different from prompt engineering alone, and where the main open problems still are.

Read the post