SHEN Haiyang

SHEN Haiyang

Peking University

GitHub RSS

😊 About Me

My research interests lie in agents based on large models (LLMs/MLMs). Specifically, I am interested in:

  • The integration of LMs with APIs (tools).
  • Retrieval augmented generation.
  • Multi-Agent Scaling.

Recently, I aim to explore the integration of LLM inference scaling with Tools.

My research group primarily focuses on system software, including machine learning systems, web systems, software engineering, and etc. However, during my second year of PhD studies, I discovered a greater interest in AI, so my current focus is primarily on AI :-).

📝 Publications

  • H. Shen*, H. Yan*, et al. “Synthetic Data for Robust and Faithful RAG Component Optimization”, arXiv (arXiv’25).
  • T. Guo*, H. Shen*, et al. “MASS: Multi-Agent Simulation Scaling for Portfolio Construction”, arXiv, (arXiv’25).
  • Q. Yang, W. B, H.Shen et al. “PixelWeb: The First Web GUI Dataset with Pixel-Wise Labels”, arXiv (arXiv’25).
  • Z. Chen, Y. Ma, H.Shen et al. “WeInfer: Unleashing the Power of WebGPU on LLM Inference in Web Browsers”, the Web Conference (WWW’25).
  • H. Shen, Y. Li et al. “Shortcutsbench: A large-scale real-world benchmark for API-based agents” the International Conference on Learning Representations (ICLR’25).
  • M. Liu, H.Shen et al. “WebAssembly for Container Runtime: Are We There Yet?” Transactions on Software Engineering Methodology (TOSEM’24).

  • H. Shen, Y. Ma et al. “Adpal: Automatic detection of troubled users in online service systems via page access logs”, IEEE International Conference on Web Services (ICWS’23).

📖 Experience

đŸ“« Correspondence