Fei Wu
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolFei Wu is a PhD-level engineer based in San Francisco with eight years of hands-on experience in distributed deep learning, cloud computing, and optimization. Skilled in TensorFlow and PyTorch development, Fei focuses on making large-scale model training and deployment more efficient—evidenced by contributions to AWS Deep Learning Containers where they integrated and optimized SageMaker model-parallel binaries across multiple PyTorch versions and enabled EFA testing. Comfortable at the intersection of research and production, Fei brings both academic rigor and pragmatic MLOps experience to engineering teams. Colleagues rely on Fei for solving performance bottlenecks in multi-node training and streamlining containerized ML workflows for the cloud.
8 years of coding experience
Stony Brook University
Github Skills (8)
pytorch10
docker10
mlops10
aws10
parallelization10
dockers10
sagemaker10
cicd9
Programming languages (5)
Github contributions (5)
aws/deep-learning-containers
Apr 2022 - Oct 2022
AWS Deep Learning Containers are pre-built Docker images that make it easier to run popular deep learning frameworks and tools on AWS.
Role in this project:
MLOps Engineer
MLOps EngineerContributions:4 reviews, 5 commits, 4 PRs in 6 months
Contributions summary:Fei's commits primarily revolve around integrating and optimizing the SageMaker model parallel (SMP) binaries for deep learning containers. They focused on releasing and updating SMP binaries for different PyTorch versions, including PT1.10, PT1.11, and PT1.12. The contributions include enabling and testing EFA (Elastic Fabric Adapter) tests, adding zero-2d, and refining the build process for various configurations, thereby contributing to the efficient deployment and execution of deep learning models on AWS infrastructure.
pytorchsagemakercontainersmxnetserving
aws-samples/mask-rcnn-tensorflow
Jul 2019 - Jan 2020
Fork of Tensorpack to make breaking performance improvements to the Mask RCNN example. Training is approximately 2x faster than the original implementation on AWS.
Contributions:135 commits, 23 PRs, 112 pushes in 5 months
maskrcnnfastermask-rcnntraining
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.