Yunxuan Xiao
Software Engineer at Google
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolYunxuan Xiao is a software engineer with nine years of experience building large-scale distributed systems and ML infrastructure, currently working on GenAI inference efficiency for Gemini at Google in San Francisco. Previously a core contributor to Ray AI Runtime at Anyscale, he implemented scalable training and tuning solutions—including a cost-optimized diffusion pre-training that saved 3x on 14,000 A100 hours—and contributes to the flagship open-source Ray project to improve debugging and result handling. His background spans systems work at NVIDIA (CUDA kernel interfaces), autonomous vehicle planning at Meituan, and language-modeling research from CMU and ByteDance, giving him a strong bridge between research and production. Known for improving developer ergonomics in complex distributed ML tooling, he combines deep systems knowledge with practical optimizations that reduce cost and operational friction.
9 years of coding experience
4 years of employment as a software developer
Master's degree Computational Data Science (System & AI), Master's degree Computational Data Science (System & AI) at Carnegie Mellon UniversityBachelor's degree Computer Science (ACM Honored Class), Bachelor's degree Computer Science (ACM Honored Class) at Shanghai Jiao Tong University
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (5)
debugging10
debug10
python10
machine-learning9
data-science8
Programming languages (7)
Github contributions (5)
ray-project/ray
Nov 2022 - Sep 2024
Ray is an AI compute engine. Ray consists of a core distributed runtime and a set of AI Libraries for accelerating ML workloads.
Role in this project:
Back-end Developer
Back-end DeveloperContributions:736 reviews, 201 PRs, 16 pushes in 1 year 10 months
Contributions summary:Yunxuan's contributions center around the development of components related to Ray Tune's internal classes. They focused on adding a representation method (`__repr__()`) for the `ResultGrid` class and enhancing the representation of the `Result` class. These changes involved modifying Python files within the `ray/tune` and `ray/air` directories, specifically addressing the display and debugging capabilities of the framework's results handling. This suggests a focus on improving the usability and debugging experience for developers using Ray Tune.
pythonconsistsruntimetensorflowserving
woshiyyya/ray
Jan 2023 - Sep 2024
Ray is a unified framework for scaling AI and Python applications. Ray consists of a core distributed runtime and a toolkit of libraries (Ray AIR) for accelerating ML workloads.
Contributions:1 review, 1 PR, 944 pushes in 1 year 7 months
raypythondata-sciencedeep-learningconsists
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.