Zhaoqi Zhu
Staff Software Engineer at Meta
California, United States
Summary
🤩
Rockstar🎓
Top SchoolZhaoqi Zhu is a Staff Software Engineer based in California with 10 years of experience specializing in C++, high-performance computing, and systems programming for production ML workloads. He has driven low-level collective communication and RDMA-backed CPU/GPU data paths at AWS for SageMaker and Trainium, and now focuses on performance optimization for production models at Meta. An Apache MXNet committer and contributor, he implemented JSON-profile aggregation and finer-grained profiler events that improved debugging and observability for a widely used deep learning engine. He combines strong academic results from USC with hands-on system and algorithm design—authoring all_reduce/all_gather/reduce_scatter variants and custom PyTorch process groups—for large-scale distributed training. Notably, he moves seamlessly between shipping production infrastructure and refining developer-facing tooling, revealing a talent for turning complex low-level optimizations into practical, debuggable systems.
10 years of coding experience
6 years of employment as a software developer
Master of Science - MS, Computer Science, 4.0/4.0, Master of Science - MS, Computer Science, 4.0/4.0 at University of Southern CaliforniaHigh School Diploma, 4.25/4.3, High School Diploma, 4.25/4.3 at Peking University High School
Chinese, English
Github Skills (7)
mxnet10
testing9
cprogramming-language9
c-language9
json9
api-design8
performance-analysis7
Programming languages (6)
Github contributions (5)
apache/mxnet
Jun 2019 - Jul 2021
Lightweight, Portable, Flexible Distributed/Mobile Deep Learning with Dynamic, Mutation-aware Dataflow Dep Scheduler; for Python, R, Julia, Scala, Go, Javascript and more
Role in this project:
ML Engineer
ML EngineerContributions:277 reviews, 148 commits, 467 PRs in 2 years 1 month
Contributions summary:Zhaoqi's primary focus involved enhancing the MXNet profiler API. They implemented features for sorting and printing aggregate profiling information in JSON format, a major addition. The user also addressed several bugs and added test cases, demonstrating a focus on improving the usability and functionality of the profiling tools. Furthermore, they refactored code to improve code style and parameter validation.
pythonschedulerdataflowmutationdata-science
Zha0q1/incubator-mxnet
Jul 2020 - Jul 2021
Lightweight, Portable, Flexible Distributed/Mobile Deep Learning with Dynamic, Mutation-aware Dataflow Dep Scheduler; for Python, R, Julia, Scala, Go, Javascript and more
Contributions:2 releases, 963 pushes, 158 branches in 11 months
pythonschedulerdataflowmutationorchestration
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.