Abstract¶
Scientific researchers need reproducible software environments for complex applications that can run across heterogeneous computing platforms. Modern open source tools, like Pixi, provide automatic reproducibility solutions for all dependencies while providing a high level interface well suited for researchers.
This tutorial will provide a practical introduction to using Pixi to easily create scientific and AI/ML environments that benefit from hardware acceleration, across multiple machines and platforms. The focus will be on CUDA applications, such as machine learning frameworks and use of CUDA Tile, as well as using pixi-build to construct bespoke CUDA enabled conda packages.
Taught at SciPy 2026 as a tutorial on Monday July 13th, 2026
SciPy Logistical Information¶
Tutorial name: Reproducible CUDA Accelerated Workflows for Scientists with Pixi
Date: 2026-07-13 (Monday)
Time: 13:30–17:30 Central
Location: Room HSEC 2-138, Health Sciences Education Center, University of Minnesota
526 Delaware Street SE, Minneapolis, MN 55455
Rough Outline¶
00:00 – 00:05 (5 min):
Personal Introductions.
00:05 – 00:15 (10 min):
Setup instructions, setup your machine for the tutorial.
00:15 – 00:30 (15 min):
Introduction to Philosophy, an overview of the philosophy behind this tutorial.
00:30 – 01:00 (30 min):
Pixi introduction, an overview of Pixi’s features and capabilities.
01:00 – 01:40 (40 min):
Pixi exercises, play around with Pixi and create a reproducible Python environment.
01:40 – 01:55 (15 min):
Break, grab a snack and stretch your legs.
01:55 – 02:35 (40 min):
Introduction to CUDA and cuTile Python, overview of CUDA and introduction to the cuTile Python library.
02:35 – 02:45 (10 min):
Break, grab a snack and stretch your legs.
02:45 – 03:15 (30 min):
Building CUDA packages from source with Pixi Build, Introduction to building CUDA packages from source with Pixi Build.
03:15 – 03:50 (35 min):
Building Package Exercises, package your own CUDA C++ and Python/cuTile code as conda packages.
03:50 – 04:00 (10 min):
Discussion over solutions to exercises.
Time for participants to start exploring their own projects with Pixi and CUDA.
General question and answer time with instructor team.
This tutorial was supported by the University of Wisconsin–Madison Data Science Institute, prefix.dev GmbH, and NVIDIA.
Acknowledgments¶
Special thanks to Tim de Jager and Wolf Vollprecht of prefix.dev GmBH and Daniel Ching of NVIDIA for their review, feedback, and contributions to this tutorial. Special thanks to Nick Hodgskin and Eniola Awowale for volunteering to serve as teaching assistants for this tutorial.
- Feickert, M., Arts, R., & Riehl, K. (2026). Reproducible CUDA Accelerated Workflows for Scientists with Pixi. Zenodo. 10.5281/ZENODO.21829174