Stanley Winata
Senior Member Of Technical Staff - AI Group at AMD
San Francisco Bay Area United States
Summary
🤩
Rockstar🎓
Top SchoolStanley Winata is a Senior Member of Technical Staff in AMD’s AI Group with five years of hands-on experience accelerating deep learning on real hardware. He holds an MS in Robotic Systems Development from Carnegie Mellon and blends robotics-rooted systems thinking with GPU compiler and performance engineering. Stanley helped bring Nod Labs into AMD and has contributed substantive ROCm backend support and optimizations to the IREE compiler, enabling AMD GPU codegen and runtime features like shared memory and descriptor sets. He also drives MLOps and benchmarking work in open-source projects such as SHARK Studio, adding practical performance runners that span PyTorch and IREE benchmarks. Based in the Bay Area, he’s comfortable shipping low-level, performance-critical components while keeping an eye on developer tooling and reproducible benchmarking.
5 years of coding experience
5 years of employment as a software developer
Ontario Secondary School Diploma, Ontario Secondary School Diploma at Bronte College of CanadaMaster of Science - MS, Robotic System Development, Master of Science - MS, Robotic System Development at Carnegie Mellon University School of Computer ScienceBachelor of Engineering (BEng), Mechanical and Aerospace Engineering, Bachelor of Engineering (BEng), Mechanical and Aerospace Engineering at Nagoya University
English, Indonesian, Chinese, Japanese, hokkien
Github Skills (20)
pytorch10
benchmark10
c-language10
python10
back-end-development10
gpu-programming10
machine-learning10
benchmarking10
roc10
mlops10
performance-optimization10
compiler10
cprogramming-language10
llvm9
mlr9
Programming languages (8)
Github contributions (5)
nod-ai/SHARK-Studio
Apr 2022 - Jan 2023
SHARK Studio -- Web UI for SHARK+IREE High Performance Machine Learning Distribution
Role in this project:
MLOps Engineer
MLOps EngineerContributions:72 reviews, 31 commits, 57 PRs in 8 months
Contributions summary:Stanley focused on enhancing the SHARK Studio project by implementing and integrating benchmarking capabilities. They added a `SharkBenchmarkRunner` class to assess model performance using different methods, including PyTorch and C benchmarks, and integrated the `iree-benchmark-module`. Furthermore, the user improved the codebase by fixing test issues on macOS and refactoring the `get_iree_module` function. The user also contributed to streamlining the build process by installing packages to the pip venv during setup.
pytorchcudaamdheterogeneousdeep-learning
iree-org/iree
May 2021 - Dec 2022
A retargetable MLIR-based machine learning compiler and runtime toolkit.
Role in this project:Back-end Developer & Performance Engineer
Back-end Developer & Performance EngineerContributions:461 reviews, 18 commits, 151 PRs in 1 year 6 months
Contributions summary:Stanley focused on developing and optimizing the ROCm backend for the IREE compiler. They implemented initial support for the ROCm HAL backend, enabling code generation and execution on AMD GPUs. Contributions include linking with ROCm device libraries, integrating shared memory support, and adding features like push constants and descriptor sets to match the CUDA HAL. Furthermore, they refactored the multi-reduction operator distribution and addressed various bugs to improve the performance and correctness of the ROCm backend.
mlirspirvvulkantensorflowcompiler
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.