Jungwook Park
AI GPU Compiler Engineer at AMD
Watford, England, United Kingdom
Summary
🤩
Rockstar🎓
Top SchoolJungwook Park is a Senior AI GPU Compiler Engineer with 7 years of focused experience building and optimizing ML compilers and GPU backends, currently contributing to Triton ROCm development at AMD. He has deep expertise in MLIR-based compiler transformations, loop pipelining, and GPU memory management, with notable open-source contributions to Triton and LLVM that improved dynamic loop handling, stream pipelining, and shared memory allocation for AMD GPUs. His background blends academic GPGPU research (PhD and postdoc work) with production engineering across Imagination Technologies, AMD, and Innosilicon, giving him a strong track record in tensor graph compilers and performance tuning. Notably, his commits reveal hands-on solutions to alignment, prefetching, and epilogue edge cases—practical hallmarks of someone who bridges low-level hardware quirks and compiler theory.
7 years of coding experience
14 years of employment as a software developer
phd, computer science, phd, computer science at Yonsei University
Github Skills (16)
compiler-optimization10
triton10
shared-memory10
c-language10
amdgpu10
cprogramming-language10
intermediate-language10
mlr10
intermediate-code10
llvm10
data-structure9
computer-engineering9
code-generation9
algorithm9
data-structures9
Programming languages (5)
Github contributions (5)
llvm/llvm-project
Dec 2020 - Apr 2025
The LLVM Project is a collection of modular and reusable compiler and toolchain technologies.
Role in this project:
Back-end Developer
Back-end DeveloperContributions:19 reviews, 8 PRs, 31 comments in 4 years 4 months
Contributions summary:Jungwook primarily focused on enhancing the MLIR (Multi-Level Intermediate Representation) dialect for the LLVM project. Their work centered on implementing and refining loop pipelining, specifically adding support for dynamic loops and addressing edge cases in the epilogue and iteration calculations. The commits include code modifications to core files related to loop transformations, demonstrating an in-depth understanding of compiler optimizations and intermediate representation manipulation. Several commits addressed issues and bugs in the pipeline logic.
compilerllvmtoolchain
triton-lang/triton
May 2024 - Jul 2026
Development repository for the Triton language and compiler
Role in this project:Backend Developer
Backend DeveloperContributions:122 reviews, 36 PRs, 144 comments in 2 years 2 months
Contributions summary:Jungwook's commits primarily focus on optimizing shared memory allocation and implementing the stream pipeliner within the Triton language and compiler. They addressed alignment issues, improved memory access patterns with local stores, and refactored the pipelining infrastructure. Furthermore, the user contributed to enabling dynamic loop peeling and improved the clustering and prefetching strategies within the stream pipeline, demonstrating a deep understanding of memory management and compiler optimization techniques for AMD GPUs.
compiler
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.