Sijie Li

prof_pic.jpg

GHC 6415

5000 Forbes Ave

Pittsburgh, PA 15213

Hello! I’m a first-year PhD student at the Language Technologies Institute, Carnegie Mellon University, advised by Yiming Yang. Before that, I received my B.S. from the Turing Class at Peking University, where I was fortunate to be advised by Di He.

My goal is to build LLM agents that operate over long-horizon machine-learning development workflows. Lately I’ve been especially drawn to decentralized multi-agent systems for long-running tasks, where progress comes from agents sharing what they learn rather than working in isolation.

To make those agents both stronger and cheaper to run, I also work on efficient model architectures and pretraining scaling laws for LLMs.

selected publications

  1. COLM
    Spend Less, Fit Better: Budget-Efficient Scaling Law Fitting via Active Experiment Selection
    Sijie Li, Shanda Li, Haowei Lin, and 3 more authors
    In Conference on Language Modeling (COLM), 2026
    Oral presentation at Sci-FM @ COLM 2026.
    arXiv
  2. ICLR
    CoMind: Towards community-driven agents for machine learning engineering
    Sijie Li, Weiwei Sun, Shanda Li, and 2 more authors
    In International Conference on Learning Representations, 2026
    arXiv
  3. EACL
    Enhancing auto-regressive chain-of-thought through loop-aligned reasoning
    Qifan Yu, Zhenyu He, Sijie Li, and 4 more authors
    In Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics, 2026
    Oral presentation.
    arXiv