Chien-Yu Lin

About me

I’m a Research Scientist in the FAIR group at Meta Superintelligence Labs. My research focuses on efficient AI through the co-design of models, algorithms, and systems, spanning model compression, GPU kernels, and training and inference systems.

I received my Ph.D. in 2025 from the Paul G. Allen School of Computer Science & Engineering at the University of Washington. I was fortunate to be advised by Luis Ceze and to work closely with Baris Kasikci and Arvind Krishnamurthy. During my time at UW, I worked on efficient RAG inference, KV-cache compression, low-bit LLM quantization, and GNN inference. During internships at Apple, I studied efficient NeRF rendering and weight sharing. Before UW, I earned my B.S. and M.S. in Electronics Engineering from National Yang Ming Chiao Tung University (formerly NCTU), where I worked on sparse CNN accelerator design under the guidance of Prof. Bo-Cheng Lai.

Beyond research, I enjoy running, cycling, tennis, mountaineering, and backcountry skiing. I lead group expeditions in the mountains and have summited and skied down all five major volcanoes in Washington State: Mount Rainier, Mount Adams, Mount Baker, Glacier Peak, and Mount St. Helens. In summer 2025, I cycled 1,000 miles from Seattle to Menlo Park; in summer 2026, I continued another 600 miles to San Diego, completing my goal of cycling the Pacific Coast from Seattle to San Diego.

News

Jan 2026	One paper (TeleRAG) got accepted to MLSys 2026 and two papers (Composer and UniQL) got accepted to ICLR 2026.
Aug 2025	Joined Meta at the MPK!
Jul 2025	Completed a 1,000-mile bike trip from Seattle to Menlo Park! It’s a journey I’ll remember for my life.

Selected publications

* means equal contribution

TeleRAG: Efficient Retrieval-Augmented Generation Inference with Lookahead Retrieval

Chien-Yu Lin^*, Keisuke Kamahori^*, Yiyu Liu^*, and 11 more authors

In Proceedings of Machine Learning and Systems (MLSys), 2026

Bib PDF

@inproceedings{lin2025telerag,
  title = {TeleRAG: Efficient Retrieval-Augmented Generation Inference with Lookahead Retrieval},
  author = {Lin, Chien-Yu and Kamahori, Keisuke and Liu, Yiyu and Shi, Xiaoxiang and Kashyap, Madhav and Shao, Rulin and Gu, Yile and Ye, Zihao and Zhu, Kan and Kadekodi, Rohan and Krishnamurthy, Arvind and Wang, Stephanie and Ceze, Luis and Kasikci, Baris},
  booktitle = {Proceedings of Machine Learning and Systems (MLSys)},
  venue_url = {https://arxiv.org/abs/2502.20969},
  year = {2026},
}

Palu: Compressing KV Cache via Low-Rank Projection

Chi-Chih Chang^*, Wei-Cheng Lin^*, Chien-Yu Lin^*, and 7 more authors

In Proceedings of International Conference on Learning Representations (ICLR), 2025

Bib PDF

@inproceedings{chang2024palu,
  author = {Chang, Chi-Chih and Lin, Wei-Cheng and Lin, Chien-Yu and Hu, Yu-Fang and Wang, Pei-Shuo and Chen, Chong-Yan and Huang, Ning-Chi and Ceze, Luis and Abdelfattah, Mohamed S. and Wu, Kai-Chiang},
  booktitle = {Proceedings of International Conference on Learning Representations (ICLR)},
  title = {Palu: Compressing KV Cache via Low-Rank Projection},
  year = {2025},
  venue_url = {https://arxiv.org/abs/2407.21118},
}

Atom: Low-Bit Quantization for Efficient and Accurate LLM Serving

Yilong Zhao, Chien-Yu Lin, Kan Zhu, and 7 more authors

In Proceedings of Machine Learning and Systems (MLSys), 2024

Bib PDF Code

@inproceedings{zhao2024atom,
  author = {Zhao, Yilong and Lin, Chien-Yu and Zhu, Kan and Ye, Zihao and Chen, Lequn and Zheng, Size and Ceze, Luis and Krishnamurthy, Arvind and Chen, Tianqi and Kasikci, Baris},
  booktitle = {Proceedings of Machine Learning and Systems (MLSys)},
  pages = {196--209},
  title = {Atom: Low-Bit Quantization for Efficient and Accurate LLM Serving},
  url = {https://proceedings.mlsys.org/paper_files/paper/2024/file/5edb57c05c81d04beb716ef1d542fe9e-Paper-Conference.pdf},
  volume = {6},
  year = {2024},
}

FastSR-NeRF: Improving NeRF Efficiency on Consumer Devices with A Simple Super-Resolution Pipeline

Chien-Yu Lin, Qichen Fu, Thomas Merth, and 2 more authors

In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), Jan 2024

Oral (Top 2.6%)

Bib PDF

@inproceedings{lin2024fastsrnerf,
  author = {Lin, Chien-Yu and Fu, Qichen and Merth, Thomas and Yang, Karren and Ranjan, Anurag},
  title = {FastSR-NeRF: Improving NeRF Efficiency on Consumer Devices with A Simple Super-Resolution Pipeline},
  booktitle = {Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)},
  month = jan,
  year = {2024},
  pages = {2482-2491},
  venue_type = {Conference},
}

SPIN: An Empirical Evaluation on Sharing Parameters of Isotropic Networks

Chien-Yu Lin^*, Anish Prabhu^*, Thomas Merth, and 4 more authors

In Proceedings the 17th European Conference on Computer Vision (ECCV), Jan 2022

Bib PDF Video Code

@inproceedings{lin2022spin,
  author = {Lin, Chien-Yu and Prabhu, Anish and Merth, Thomas and Mehta, Sachin and Ranjan, Anurag and Horton, Maxwell and Rastegari, Mohammad},
  title = {SPIN: An Empirical Evaluation on Sharing Parameters of Isotropic Networks},
  booktitle = {Proceedings the 17th European Conference on Computer Vision (ECCV)},
  year = {2022},
  venue_type = {Conference},
}

Supporting compressed-sparse activations and weights on SIMD-like accelerator for sparse convolutional neural networks

Chien-Yu Lin, and Bo-Cheng Lai

In Proceedings of the 23rd Asia and South Pacific Design Automation Conference (ASP-DAC), Jan 2018

Bib PDF Slides

@inproceedings{lin2018supporting,
  author = {Lin, Chien-Yu and Lai, Bo-Cheng},
  title = {Supporting compressed-sparse activations and weights on SIMD-like accelerator for sparse convolutional neural networks},
  booktitle = {Proceedings of the 23rd Asia and South Pacific Design Automation Conference (ASP-DAC)},
  year = {2018},
}