Tiancheng Hu

Tiancheng Hu

Third-year Ph.D. student at Peking University

About me

I’m a third-year Ph.D. student at Peking University (PKU), advised by Prof. Tao Xie.

Research

My research focuses on GPU systems and efficient AI infrastructure, with an emphasis on resource management, heterogeneous computing, and communication for AI workloads.

  • GPU scheduling and sharing: microsecond-scale preemption for SLO-oriented GPU sharing (Hummingbird).
  • Heterogeneous GPU computing: kernel-granularity disaggregation for efficient large-model inference (FluidGPU).
  • Communication for AI systems: fabric-native communication for production supernodes (StrataCL).
  • Efficient LLM inference: position-independent KV caching and reasoning-aware sparse attention (EPIC, RaaS, and Lil).

My earlier work explored fine-grained code clone detection in software engineering.

If you are interested in collaborating with me, please feel free to contact me.

Email (tiancheng.hu@stu.pku.edu.cn)

Internships

  • Huawei, Noah’s Ark Lab. · July 2025 – June 2026
    CM384 memory management and communication.

  • Ant Group · February 2024 – June 2025
    GPU sharing.

Publications

See my Google Scholar profile for the full list and citation information.

Talks