👨‍💻 About Me

I’m currently a first-year Ph.D. student in Computer Science at Nanyang Technological University (NTU), fortunate to be advised by Prof. Wenya Wang. Prior to this, I received my B.Eng. in Software Engineering from Tianjin University (TJU) and M.Eng. in Computer Science from Tsinghua University (THU).

Along the way, I’ve been fortunate to work with some amazing researchers and mentors. I worked as a research intern at University of Illinois Urbana-Champaign (UIUC) with Prof. Jiaxuan You, which turned out to be one of the most rewarding and enjoyable experiences I’ve had. I also interned as a research assistant at The Hong Kong University of Science and Technology (HKUST-GZ), where I had the pleasure of collaborating with Prof. Chengwei Qin. In addition, I spent over a year as a research intern at Microsoft Research Asia (MSRA).

Research Interests: My early research mainly focused on Graph Data Mining, and you may find some of my representative work here: TFE-GNN@WWW'23, TGSL@CIKM'23, and MH-Net@AAAI'25.

🔍 Currently, I’m focusing on LLM/Agent Memory & Routers and their synergy with Graphs. Some of my recent works include: GoR@ACL'25 Main , Router-R1@NeurIPS'25, FusionFactory, LLMRouter, MemSkill, and BudgetMem.

📬 Feel free to reach out via wazhz14 [AT] gmail [DOT] com — I’m always open to discussion and collaboration!

🔥 News

[2026.02]  🚀🚀 Introducing BudgetMem, a runtime agent memory framework for explicit performance–cost control. BudgetMem equips each memory module with Low/Mid/High budget tiers and learns query-aware budget-tier routing to deliver controllable on-demand memory extraction and clear performance–cost trade-offs. We hope BudgetMem supports future research on controllable runtime agent memory, and we welcome feedback and community contributions.

[2026.02]  🚀🚀 Released MemSkill, a framework for learning and evolving memory skills for LLM agents. MemSkill shifts agent memory to span-level, skill-conditioned generation, enabling reusable and self-evolving memory behaviors through a closed-loop of skill usage and skill evolution. We hope MemSkill helps advance more adaptive, self-evolving agent memory systems and welcome feedback and contributions from the community!

[2025.12]  🚀🚀 Open-sourced LLMRouter, a unified and extensible framework for training, evaluating, and benchmarking single-round, multi-round, agentic, and personalized LLM routers. LLMRouter is designed to reduce duplicated engineering effort, enable fair comparison across routing methods, and foster community-driven development. We warmly welcome contributions and integrations from the community!

[2025.09]  🎉🎉 Router-R1 was accepted by NeurIPS 2025. See you in San Diego!

[2025.09]  🎉🎉 AcademicEval was accepted by TMLR.

[2025.07]  🎉🎉 Introducing FusionFactory: a systematic framework for fusing LLM capabilities from routing data across query, thought, and model levels.

[2025.06]  🎉🎉 Released Router-R1: an RL-driven, multi-round LLM router that balances accuracy, cost, and efficiency.

[2025.05]  🎉🎉 One paper (GoR) was accepted by ACL 2025 Main. See you in Vienna!

[2024.12]  🎉🎉 One paper (MH-Net) was accepted by AAAI 2025.

📝 Selected Publications [Full List]

(* denotes equal contribution)

Preprints

  • Rethinking Memory Mechanisms of Foundation Agents in the Second Half: A Survey
    Wei-Chieh Huang, Weizhi Zhang, Yueqing Liang, Yuanchen Bei, Yankai Chen, Tao Feng, Xinyu Pan, Zhen Tan, Yu Wang, Tianxin Wei, Shanglin Wu, Ruiyao Xu, Liangwei Yang, Rui Yang, Wooseong Yang, Chin-Yuan Yeh, Hanrong Zhang, Haozhen Zhang, Siqi Zhu, Henry Peng Zou, Wanjia Zhao, Song Wang, Wujiang Xu, Zixuan Ke, Zheng Hui, Dawei Li, Yaozu Wu, Langzhou He, Chen Wang, Xiongxiao Xu, Baixiang Huang, Juntao Tan, Shelby Heinecke, Huan Wang, Caiming Xiong, Ahmed A. Metwally, Jun Yan, Chen-Yu Lee, Hanqing Zeng, Yinglong Xia, Xiaokai Wei, Ali Payani, Yu Wang, Haitong Ma, Wenya Wang, Chengguang Wang, Yu Zhang, Xin Wang, Yongfeng Zhang, Jiaxuan You, Hanghang Tong, Xiao Luo, Xue Liu, Yizhou Sun, Wei Wang, Julian McAuley, James Zou, Jiawei Han, Philip S. Yu, and Kai Shu
    arXiv 2602.06052, 2026
    [arXiv]   [Code]

  • Learning Query-Aware Budget-Tier Routing for Runtime Agent Memory
    Haozhen Zhang*, Haodong Yue*, Tao Feng, Quanyu Long, Jianzhu Bao, Bowen Jin, Weizhi Zhang, Xiao Li, Jiaxuan You, Chengwei Qin, and Wenya Wang
    arXiv 2602.06025, 2026
    [arXiv]   [Code]

  • MemSkill: Learning and Evolving Memory Skills for Self-Evolving Agents
    Haozhen Zhang, Quanyu Long, Jianzhu Bao, Tao Feng, Weizhi Zhang, Haodong Yue, and Wenya Wang
    arXiv 2602.02474, 2026
    [arXiv]   [Code]

  • FusionFactory: Fusing LLM Capabilities with Multi-LLM Log Data
    Tao Feng*, Haozhen Zhang*, Zijie Lei, Pengrui Han, Mostofa Patwary, Mohammad Shoeybi, Bryan Catanzaro, and Jiaxuan You
    arXiv 2507.10540, 2025
    [arXiv]   [Code]

Journal

  • AcademicEval: Live Long-Context LLM Benchmark
    Haozhen Zhang, Tao Feng, Pengrui Han, and Jiaxuan You
    Transactions on Machine Learning Research (TMLR), 2025
    [arXiv]   [Code]

Conference

  • Router-R1: Teaching LLMs Multi-Round Routing and Aggregation via Reinforcement Learning
    Haozhen Zhang, Tao Feng, and Jiaxuan You
    The Thirty-Ninth Annual Conference on Neural Information Processing Systems (NeurIPS), 2025
    [arXiv]   [Code]     ⭐ 100+ GitHub Stars

  • Graph of Records: Boosting Retrieval Augmented Generation for Long-context Summarization with Graphs
    Haozhen Zhang, Tao Feng, and Jiaxuan You
    The 63rd Annual Meeting of the Association for Computational Linguistics (ACL Main), 2025
    [arXiv]   [Code]

Master’s-stage work (click to expand)
  • Revolutionizing Encrypted Traffic Classification with MH-Net: A Multi-View Heterogeneous Graph Model
    Haozhen Zhang*, Haodong Yue*, Xi Xiao, Le Yu, Qing Li, Zhen Ling, and Ye Zhang
    The 39th Annual AAAI Conference on Artificial Intelligence (AAAI), 2025
    [arXiv]   [Code]

  • Time-aware Graph Structure Learning via Sequence Prediction on Temporal Graphs
    Haozhen Zhang, Xueting Han, Xi Xiao, and Jing Bai
    International Conference on Information and Knowledge Management (CIKM), 2023
    [arXiv]   [Code]

  • TFE-GNN: A Temporal Fusion Encoder Using Graph Neural Networks for Fine-grained Encrypted Traffic Classification
    Haozhen Zhang, Le Yu, Xi Xiao, Qing Li, Francesco Mercaldo, Xiapu Luo, and Qixu Liu
    The Web Conference (WWW), 2023
    [arXiv]   [Code]        🏆 100+ Citations     ⭐ 100+ GitHub Stars

🧩 Open-source Projects

  • LLMRouter: A Unified Framework for LLM Routing
    Tao Feng, Haozhen Zhang, Zijie Lei, Haodong Yue, Chongshan Lin, and Jiaxuan You
    A unified and extensible framework for training, evaluating, and benchmarking single-round, multi-round, agentic, and personalized LLM routers. LLMRouter aims to reduce duplicated engineering effort, enable fair comparison across routing methods, and foster community-driven development of LLM routing systems.
    [GitHub]   [Project Page]     ⭐ 1.3K+ GitHub Stars

📖 Educations

  • Present: Ph.D. Student in Computer Science, Nanyang Technological University (NTU).

  • M.Eng. in Computer Science, Tsinghua University (THU).

  • B.Eng. in Software Engineering, Tianjin University (TJU).

💻 Professional Experience

  • 2025.05 - 2025.08, The Hong Kong University of Science and Technology (HKUST-GZ) - Research Assistant.

  • 2024.06 - 2025.06, University of Illinois Urbana-Champaign (UIUC) - Affiliated Researcher.

  • 2022.11 - 2024.02, Microsoft Research Asia (MSRA) - Research Intern.

🤝 Professional Service

  • Reviewer:

    • AAAI 2026

    • NeurIPs 2025

    • KDD 2024, ICLR 2024

🎯 Miscellaneous

I really enjoy playing badminton🏸! I was on the university or departmental team during both undergrad and grad school.

I also trained in table tennis🏓 as a kid — though it’s been quite a few years since I last picked up a paddle…😅

✉️ Contact

Email: wazhz14 [AT] gmail [DOT] com

Feel free to reach out — I’m always open to discussion and collaboration!