Ziming Miao
Senior Researcher at Microsoft Research Asia
Haidian District, Beijing, China
Summary
🤩
Rockstar🎓
Top SchoolZiming Miao is a Senior Researcher at Microsoft Research Asia with nine years of experience building and optimizing AI infrastructure and model compilers. He blends research rigor with production engineering from prior SDE work at Microsoft, focusing on scalable cluster management, GPU-aware DevOps, and DNN compilation. His open-source contributions include performance and deployment improvements to Microsoft PAI and feature additions to NNFusion—enabling large BERT and TorchScript workflows—showing a knack for bridging model formats to high-performance runtimes. Trained in signal and information processing (UCAS) with a strong electronics foundation (USTC), he excels at turning complex ML tooling challenges into automated, testable solutions. An often-overlooked strength is his ability to accelerate end-to-end testing and cleanup automation in cluster environments, reducing friction for large-scale experiments.
9 years of coding experience
3 years of employment as a software developer
Master's degree, Signal and Information Processing, Master's degree, Signal and Information Processing at University of Chinese Academy of SciencesBachelor's degree, Electronic Science and Technology, Bachelor's degree, Electronic Science and Technology at University of Science and Technology of China
Github Skills (21)
pytorch10
kubernetes10
docker10
onnx10
dockers10
cicd10
build-automation10
kubernetes-pods10
scripting9
torchscript9
machine-learning9
neural-network9
shell9
script9
sh9
Programming languages (6)
Github contributions (5)
microsoft/nnfusion
Sep 2020 - Jan 2023
A flexible and efficient deep neural network (DNN) compiler that generates high-performance executable from a DNN model description.
Role in this project:
ML Engineer
ML EngineerContributions:79 reviews, 82 commits, 118 PRs in 2 years 4 months
Contributions summary:Ziming primarily contributes to the development and enhancement of a deep neural network (DNN) compiler. Their work includes adding support for specific ONNX models, such as a large BERT model, and integrating functionalities like support for softmax weights in the ONNX frontend. Additionally, the user is involved in enabling TorchScript support, suggesting a focus on improving the compiler's ability to handle different model formats and integrate with other AI toolchains. The commits indicate improvements in ONNX frontend op type checking.
tvmdeep-learningdeep-neural-networkcompilerneural-network
microsoft/pai
Jul 2018 - Sep 2019
Resource scheduling and cluster management for AI
Role in this project:DevOps Engineer
DevOps EngineerContributions:1 release, 1 review, 132 commits in 1 year 1 month
Contributions summary:Ziming primarily focused on enhancing the build and deployment processes of the `microsoft/pai` repository. Their work included modifying build scripts to incorporate GPU-related patches, adjusting the docker executor configuration, and adding scripts to clean up running jobs during node deletion. They also made changes to the end-to-end test environment to accelerate testing. The user's contributions show strong involvement in automating infrastructure-related tasks.
cluster-managementmodel-trainingchainertensorflowcloud
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.