Rafał Banaś
Deep Learning Engineer at NVIDIA
Warsaw, Masovian Voivodeship, Poland
Summary
🤩
Rockstar🎓
Top SchoolRafał Banaś is a Deep Learning Engineer at NVIDIA with eight years of software engineering experience and a strong foundation in computer science from the University of Warsaw. He focuses on GPU-accelerated data processing and has made notable contributions to NVIDIA DALI—adding multi-input/output Python function operators, a TorchPythonFunction for PyTorch integration, race-condition fixes, and broadening operator support for diverse data types and layouts. Comfortable across back-end and ML engineering, he blends low-level performance optimizations with practical tooling to speed up model training and inference. Prior roles include a Java development stint at Viacom and an internship that led to a full-time position at NVIDIA, highlighting a track record of turning research-grade ideas into production-ready code. An interesting detail: he works at the intersection of Python operator design and GPU kernels, a niche that combines pedagogical clarity for users with high-performance systems thinking.
8 years of coding experience
Bachelor of Science - BS, Computer Science, Bachelor of Science - BS, Computer Science at Uniwersytet Warszawski
Github Skills (12)
data-preprocessing10
pytorch10
machine-learning10
deep-learning10
gpu10
python10
data-engineering10
cupy9
c-language8
cprogramming-language8
tensorflow7
computer-vision7
Programming languages (6)
Github contributions (5)
NVIDIA/DALI
Apr 2019 - Nov 2022
A GPU-accelerated library containing highly optimized building blocks and an execution engine for data processing to accelerate deep learning training and inference applications.
Role in this project:
Back-end & ML Engineer
Back-end & ML EngineerContributions:12 releases, 795 reviews, 101 commits in 3 years 7 months
Contributions summary:Rafał primarily worked on extending and improving the Python-based DALI library, a GPU-accelerated data processing library for deep learning. Their contributions include the development of Python function operators to handle multiple inputs and outputs, and optimizations for efficient data handling. The user also implemented features such as the TorchPythonFunction operator, facilitating integration with PyTorch, and addressing a race condition in Python operator implementations. Furthermore, they worked on refactoring and improvements to image processing operators by incorporating new kernels, and extended support to various data types and layouts within the ExternalSource operator, focusing on data loading and pre-processing.
deep-learninggpuinferencefast-data-pipelineimage-augmentation
banasraf/udp-radio
Apr 2018 - Nov 2018
Contributions:31 commits, 21 pushes, 1 branch in 6 months
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.