Yi Zhu
Senior RESEARCH SDE at Microsoft
Haidian District, Beijing, China
Summary
👤
Senior🎓
Top SchoolYi Zhu is a Senior Research SDE at Microsoft with nine years of experience building distributed training and inference systems for deep learning, with deep expertise in compilers and programming languages. He has hands-on experience improving I/O and persistence in the OneFlow framework, contributing backend fixes and Hadoop-related filesystem robustness for a production-scale DL project. His background includes internships at Microsoft Research Asia and OneFlow, and a Master's from Tsinghua, reflecting a strong bridge between research and industrial deployment. Known as honest, modest, and reliable, he favors pragmatic improvements that enhance scalability and fault tolerance rather than flashy features. He also brings experience building distributed pipelines on large-scale platforms (Cosmos) and teaching computational geometry, underscoring both systems depth and mentorship ability. Based in Haidian, Beijing, Yi focuses on making deep learning infrastructure more efficient and resilient in real-world environments.
9 years of coding experience
Master's degree, Computer Software Engineering, Master's degree, Computer Software Engineering at Tsinghua UniversityBachelor's degree, Computer Software Engineering, Bachelor's degree, Computer Software Engineering at Beijing Institute of Technology
Github Skills (12)
debugging10
debug10
filesystem10
c-language10
cprogramming-language10
hadoop9
deep-learning9
version-control9
git9
build-system8
build-tools8
build-engine8
Programming languages (4)
Github contributions (5)
Oneflow-Inc/oneflow
Sep 2017 - Nov 2018
OneFlow is a deep learning framework designed to be user-friendly, scalable and efficient.
Role in this project:
Back-end Developer
Back-end DeveloperContributions:172 commits in 1 year 2 months
Contributions summary:Yi's contributions primarily revolve around modifying and enhancing the file system interfaces and related functionalities within the OneFlow deep learning framework. These changes involve updating interfaces, fixing bugs in file system implementations (specifically related to Hadoop), and addressing merge conflicts. The commits demonstrate work on core persistence functionalities, which are crucial for data handling in the context of a deep learning framework. These modifications suggest a focus on improving the framework's I/O capabilities and error handling.
caffe2tensorflowframework-learningchainerdeep-learning-framework
microsoft/nnscaler
Sep 2024 - Mar 2025
nnScaler: Compiling DNN models for Parallel Training
Contributions:11 reviews, 3 PRs, 1 push in 5 months
compilerdeep-learningdistributed-trainingllmmachine-learning
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.