Xiaozhe Yao
Scholar

Xiaozhe Yao

Google Scholar ID: Bhgm1tQAAAAJ
ETH Zurich
Machine Learning SystemsMachine LearningLLMs
Homepage↗Google Scholar↗
Citations & Impact
All-time
Citations
446
 
H-index
8
 
i10-index
8
 
Publications
19
 
Co-authors
9
list available
Contact
Emailaskxzyao@gmail.comGitHubOpen ↗LinkedInOpen ↗
Publications
6 items
Taming the Long-Tail: Efficient Reasoning RL Training with Adaptive Drafter
2025
Cited
0
Mixtera: A Data Plane for Foundation Model Training
2025
Cited
0
ThunderServe: High-performance and Cost-efficient LLM Serving in Cloud Environments
2025
Cited
0
Demystifying Cost-Efficiency in LLM Serving over Heterogeneous GPUs
2025
Cited
0
Aurora-M: Open Source Continual Pre-training for Multilingual Language and Code
2024
Cited
5
DeltaZip: Efficient Serving of Multiple Full-Model-Tuned LLMs
2023
Cited
6
Resume (English only)
Academic Achievements
  • Involved in building Swiss AI Serving
  • Proposed DeltaZip for serving multiple full-model-tuned LLMs efficiently
  • Provided an information-theoretic perspective on why delta compression works
  • Contributed to the Vision Debugging Benchmark at DataPerf
  • Built a neural network from scratch
Co-authors
9 total
Ce Zhang
Ce Zhang
Together AI; University of Chicago
Ana Klimovic
Ana Klimovic
ETH Zurich
Co-author 3
Co-author 3
Binhang Yuan（袁彬航）
Binhang Yuan（袁彬航）
Hong Kong University of Science and Technology
Beidi Chen
Beidi Chen
Carnegie Mellon University
Sanmi Koyejo
Sanmi Koyejo
Assistant Professor, Stanford University
Co-author 7
Co-author 7
Cedric Renggli
Cedric Renggli
Apple