Won Park
Seoul, South Korea
Summary
🤩
Rockstar🎓
Top SchoolWon Park is an experienced open-source developer and ML engineer with 23 years in software and research, based in Seoul. Trained as a physical chemist (PhD) at Sungkyunkwan University, he transitioned into software engineering and served as a Senior Research Engineer there, blending scientific rigor with practical coding. His contributions span full-stack UI work (stable-diffusion-webui-forge), deep internals of ML tooling (bitsandbytes k-bit quantization and CUDA build fixes), and core backend libraries (Parsedown PHP parser). He’s comfortable debugging cross-platform build and CUDA detection issues, enhancing model tooling like LoRA extraction with SVD, and shipping both JavaScript and Python fixes. Notably, his work on bitsandbytes tackled tricky Windows/CUDA compatibility problems that many ML engineers encounter when deploying quantized models. This mix of academic depth and hands-on open-source impact makes him adept at bridging research, model optimization, and production engineering.
23 years of coding experience
6 years of employment as a software developer
박사, Physical Chemistry, 박사, Physical Chemistry at 성균관대학교
Stackoverflow
115reputation
3kreached
3answers
0questions
Github Skills (42)
markdown10
parserator10
pytorch10
parser10
strings10
python10
decomposition10
text-manipulation10
machine-learning10
stable-diffusion10
php10
markdown-it10
parse10
model-optimization10
lora10
Programming languages (13)
Github contributions (5)
bitsandbytes-foundation/bitsandbytes
Nov 2023 - Feb 2024
Accessible large language models via k-bit quantization for PyTorch.
Role in this project:
ML Engineer
ML EngineerContributions:26 reviews, 11 PRs, 68 comments in 2 months
Contributions summary:Won contributed significantly to the `bitsandbytes` library, which focuses on k-bit quantization for PyTorch. Their work involved debugging and fixing library detection issues, particularly related to Windows compatibility and CUDA setup. They also modified and extended the build process, ensuring compatibility across different platforms and CUDA versions. The contributions also include the adjustment of CUDA runtime library detection for different operating systems, indicating a deep understanding of the underlying infrastructure and compilation processes.
cudapytorch8-bitgpupruning
lllyasviel/stable-diffusion-webui-forge
Jul 2024 - Oct 2024
Role in this project:Full-stack Developer
Full-stack DeveloperContributions:1 review, 2 PRs, 6 comments in 2 months
Contributions summary:Won contributed to both the frontend and backend of the project. They implemented features for the XYZ grid, fixed bugs related to parameter parsing and image preview display, and supported the webui.settings.bat functionality. Furthermore, they addressed an indexing error in the image processing module. The user also modified both JavaScript and Python code in UI related modules.
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.