Szymon Ożóg
Member Of Engineering Staff at Poolside
Silesian Voivodeship, Poland
Summary
🤩
Rockstar🎓
Top SchoolSzymon Ożóg is a software engineer with seven years of experience bridging game development and machine learning, currently serving as Member of Engineering Staff at Poolside after a stint as Senior AI Inference Engineer at Aleph Alpha. He combines production game programming expertise from Carbon Studio with hands-on ML systems work—contributing to high-throughput LLM tooling such as vllm (GGUF support and quantized model optimizations) and low-level accelerators in tinygrad (CUDA/PTX and Triton support). Comfortable across full-stack and low-level performance engineering, he focuses on model loading, memory efficiency, and inference robustness. Based in Silesia, Poland, he holds a Master’s in Computer Science and brings a practical, systems-minded approach that surfaces in both game engines and ML inference pipelines.
7 years of coding experience
Master's degree, Computer Science, Master's degree, Computer Science at The Silesian University of Technology
Stackoverflow
466reputation
18kreached
25answers
24questions
Github Skills (22)
pytorch10
gf10
python10
gfs10
machine-learning10
llm10
triton10
ptx10
quantization10
cuda10
gff10
deep-learning9
transformer8
cprogramming-language8
c-language8
Programming languages (4)
Github contributions (5)
tinygrad/tinygrad
Jul 2023 - Jul 2024
You like pytorch? You like micrograd? You love tinygrad! ❤️
Role in this project:
Full-stack Developer
Full-stack DeveloperContributions:40 reviews, 44 PRs, 87 comments in 1 year
Contributions summary:Szymon contributed significantly to the `tinygrad/tinygrad` repository, focusing on enhancing and maintaining the project's functionality. Their work involved debugging and improving the HIP error messages, making substantial updates to the codebase, and implementing features to support Triton compilation. They were also responsible for enabling, testing, and cleaning up the Triton project. The user demonstrated strong skills in the CUDA and PTX backends.
deep-learningpytorchmicrograd
vllm-project/vllm
Feb 2025 - Apr 2025
A high-throughput and memory-efficient inference and serving engine for LLMs
Role in this project:ML Engineer
ML EngineerContributions:14 reviews, 10 PRs, 40 comments in 1 month
Contributions summary:Szymon primarily contributes to the integration and improvement of GGUF (GGML Unified Format) support within the VLLM project. Their commits involve bug fixes related to GGUF weight initialization, the addition of new GGUF kernels, and implementation of GGUF support for Deepseek models. These changes demonstrate a focus on expanding the project's capabilities for quantized models and optimizing model loading processes. The user has also fixed issues related to chunked prefill operations.
amdcudadeepseekgpthpu
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.