Senior Machine Learning Compiler Engineer (IREE) at AMD
Seattle, Washington, United States
Join Prog.AI to see contacts
Join Prog.AI to see contacts
Summary
🤩
Rockstar
🎓
Top School
Krzysztof Drewniak is a Senior Machine Learning Compiler Engineer with 15 years of experience specializing in MLIR/LLVM-based code generation for machine learning and high-performance computing. Currently at AMD, he drives compiler support for new GPU hardware (RDNA4, MI-series) and leads upstream MLIR/AMDGPU backend work that tangibly improves performance and correctness. His open-source contributions include significant AMDGPU and IREE enhancements—optimizing ROCm translations, adding support for new WMMAR4 targets, and improving buffer and bounds handling in LLVM. He combines deep compiler backend expertise with practical system-level fixes (e.g., eliminating buffer-load inefficiencies that yielded measurable speedups) and a strong interest in programming languages and accessibility. A longtime Rust enthusiast, he also has roots in language and tooling work from SBCL to window manager projects, reflecting a rare blend of low-level compiler engineering and broader software craftsmanship.
15 years of coding experience
4 years of employment as a software developer
Master of Science - MS Computer Science, Master of Science - MS Computer Science at University of Washington
Bachelor’s Degree Computer Science and Mathematics, Bachelor’s Degree Computer Science and Mathematics at The University of Texas at Austin
TAMS diploma (equivalent to high school diploma) Texas Academy of Mathematics and Science, TAMS diploma (equivalent to high school diploma) Texas Academy of Mathematics and Science at University of North Texas
The LLVM Project is a collection of modular and reusable compiler and toolchain technologies.
Role in this project:
Back-end Developer
Contributions:450 reviews, 3 commits, 208 PRs in 10 days
Contributions summary:Krzysztof primarily contributed to the LLVM project by modifying and extending the code related to AMDGPU backend, focusing on areas like buffer fat pointers and vector operations. They improved the handling of GPU bounds and implemented value bounds interfaces. The contributions include adding and improving the code for various operations like memcpy, and extending buffer atomic operations for new data types. These contributions enhance the performance and capabilities of the LLVM compiler for AMDGPU architectures.
A retargetable MLIR-based machine learning compiler and runtime toolkit.
Role in this project:
Back-end Developer & Compiler Engineer
Contributions:79 reviews, 61 PRs, 107 pushes in 10 months
Contributions summary:Krzysztof primarily focused on improving the ROCm-specific LLVM translations within the IREE compiler. Their work involved optimizing the compilation process for AMD GPUs by leveraging upstream translations, correctly parsing chipset versions, and adding upper bound annotations for dispatch IDs. The user contributed to enhancements in code generation, including optimizing integer arithmetic and integrating affine transformations to enhance code efficiency. They also modified the HAL and LLVM interfaces, and added support for new AMDGPU WMMAR4 targets to extend the capabilities of the compiler.
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.
Request Free Trial
Krzysztof Drewniak - Senior Machine Learning Compiler Engineer (IREE) at AMD