Mehdi Mirzazadeh
Senior Staff Software Engineer at Google
San Francisco Bay Area United States
Summary
🤩
Rockstar🎓
Top SchoolMehdi Mirzazadeh is a Senior Staff Software Engineer based in the San Francisco Bay Area with a strong foundation in distributed systems and large-scale ML infrastructure. With over a decade of experience spanning roles at Google and Facebook, he specializes in building robust, high-performance pipelines and resolving complex memory and dependency issues in production systems. His open-source contributions to Facebook Research’s fairscale demonstrate deep expertise in PyTorch internals, automatic graph generation, and preparing distributed pipelines for new framework versions. Mehdi pairs a PhD-level academic background from the University of Waterloo with hands-on engineering leadership, enabling him to translate research insights into practical, production-ready solutions. Colleagues know him for tackling thorny infrastructure problems that bridge model training efficiency and reliable deployment.
5 years of coding experience
10 years of employment as a software developer
B.Sc., Computer Engineering, B.Sc., Computer Engineering at Sharif University of TechnologyM.Sc., Computer Science, M.Sc., Computer Science at University of Waterloo
Github Skills (12)
pytorch10
parallel-processing10
machine-learning10
deep-learning10
python10
pipeline10
distributed-computing10
graphing9
rpc9
grapher9
graph9
testing8
Programming languages (2)
Github contributions (5)
facebookresearch/fairscale
Apr 2021 - Jun 2021
PyTorch extensions for high performance and large scale training.
Role in this project:
ML Engineer
ML EngineerContributions:39 reviews, 7 commits, 22 PRs in 2 months
Contributions summary:Mehdi primarily focused on developing and refining the distributed pipeline infrastructure for PyTorch models within the fairscale library. Their contributions involved implementing features for automatic graph generation, improving the handling of dependencies between pipeline stages, and addressing memory leaks within the distributed processing framework. The user's work also included optimizing the partition handler and preparing the pipeline for compatibility with newer versions of PyTorch, indicating a deep understanding of the library's internal workings and its application in large-scale machine learning.
pytorchdeep-learningmachine-learningscaletraining
pbelevich/pipeline_experiments
Apr 2021 - Jul 2021
PyTorch pipeline parallelism experiments
Contributions:2 reviews, 9 commits, 1 PR in 2 months
pytorchpipelinemachine-learningparallelismpipeline-parallelism
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.