Wentao Guo

CS PhD student @ Princeton

Princeton University

Biography

Wentao Guo is a CS PhD student at Princeton University. His research advisor is Prof. Tri Dao. He obtained his both master’s and bachelor’s degrees in Computer Science from Cornell University. He mainly works on the intersection of hardware efficiency and ML model architecture design.

Interests

LLM efficiency
Efficient ML algorithms
Machine Learning Systems

Education

Ph.D. in Computer Science
Princeton University
MEng. in Computer Science, 2023
Cornell University, GPA 3.993
B.S. in Computer Science with Honors, 2022
Cornell University, GPA 3.890

Recent Papers

Quickly discover relevant content by filtering papers.

Wentao Guo, Mayank Mishra, Xinle Cheng, Ion Stoica, Tri Dao (2025). SonicMoE: Accelerating MoE with IO and Tile-aware Optimizations. arXiv.

PDF Cite

Wentao Guo, Jikai Long, Yimeng Zeng, Zirui Liu, Xinyu Yang, Yide Ran, Jacob R. Gardner, Osbert Bastani, Christopher De Sa, Xiaodong Yu, Beidi Chen, Zhaozhuo Xu (2025). Zeroth-Order Fine-Tuning of LLMs with Transferable Static Sparsity. ICLR'25..

PDF Cite

Wentao Guo, Andrew Wang, Bradon Thymes, Thorsten Joachims (2024). Ranking with Slot Constraints. KDD'24.

PDF Cite Code Poster DOI

A. Feder Cooper, Wentao Guo, Khiem Pham, Tiancheng Yuan, Charlie F. Ruan, Yucheng Lu, Christopher De Sa (2023). Coordinating Distributed Example Orders for Provably Accelerated Training. In NeurIPS'23.

PDF Cite Code Poster

Yucheng Lu, Wentao Guo, Christopher De Sa (2022). GraB: Finding Provably Better Data Permutations than Random Reshuffling. In NeurIPS'22.

PDF Cite Poster

See all publications

Prior Research Experience

Research Assistant

Prof. Beidi Chen, Carnegie Mellon University

June 2023 – May 2024 Remote

I worked on efficient sparse fine-tuning of LLMs with zeroth-order optimization methods.

Research Assistant

Prof. Christopher De Sa, Cornell University

June 2021 – May 2023 Ithaca, NY

I worked on (distributed) example orderings that produced the CD-GraB and GraB paper, and efficient high-precision floating-point arithmetic for hyperbolic representation learning as MCTensor and HTorch.

Research Assistant

Prof. Thorsten Joachims, Cornell University

June 2022 – February 2023 Ithaca, NY

I worked on the ranking with slot constraints problem.