I'm Huanyu Wang (王焕宇), a Ph.D. student at Carnegie Mellon University, advised by Prof. Beidi Chen at InfiniAI Lab.

I received dual degrees in Computer Science (B.E., IEEE Honor Class) and Applied Mathematics (B.S.) from Shanghai Jiao Tong University in 2026, where I was fortunate to work with Prof. Zhouhan Lin on efficient vision-language models.

Outside of research, I enjoy basketball, Japanese anime, and traveling. I also live with a lovely cat named Mao-na.

Mao-na
Mao-na (猫娜)

Research Interests

  • Efficient Inference Token compression and KV cache optimization for efficient LLM/VLM serving.
  • Decentralized Systems Cross-region collaboration between heterogeneous AI services.
  • ML Systems & HPC Kernel- and system-level optimization across modern hardware.

News

🎉 Fourier Compressor accepted to ECCV 2026!
🎓 Graduated from SJTU with dual degrees in CS and Math!
📄 Released WWW.Serve, a decentralized framework for collaborative multi-LLM serving.
📄 Released DynaKV, a token-wise adaptive KV cache compression method.
📄 Released Fourier Compressor, a frequency-domain VLM token compression approach.

Selected Publications

  • Fourier Compressor teaser
    Fourier Compressor: Frequency-Domain Visual Token Compression for Vision-Language Models
    Huanyu Wang, Jushi Kai, Haoli Bai, Lu Hou, Bo Jiang, Ziwei He, Zhouhan Lin
    ECCV 2026 [Paper] [Code]
  • WWW.Serve teaser
    WWW.Serve: Interconnecting Global LLM Services through Decentralization
    Huanyu Wang, Ziyu Xia, Zhuoming Chen, Beidi Chen
    Preprint, 2026 [Paper] [Code] [Blog]

See full publication list →

Education

CMU Logo
Carnegie Mellon University
Ph.D. in Electrical and Computer Engineering
Aug. 2026 - Present
SJTU Logo
Shanghai Jiao Tong University
B.E. in Computer Science and Technology (IEEE Honor Class)
B.S. in Mathematics and Applied Mathematics (Dual degree)
Sept. 2022 - Jun. 2026