Get Involved
There are many ways to contribute to the Fedora AI/ML SIG. You do not need to be a packager to participate. The SIG welcomes contributors of all backgrounds and skill levels.
To get started, introduce yourself in our Matrix room and let us know what interests you. You can also browse our ticket tracker for open tasks.
|
A Fedora account is required to access Fedora Infrastructure. It only takes a minute to set up. |
RPM packaging
The SIG’s core technical work is packaging AI/ML libraries for Fedora. This includes the ROCm compute stack, PyTorch, llama-cpp, Ollama, the HuggingFace ecosystem, and more.
If you are already a Fedora packager or want to become one, this is the most direct way to contribute. The SIG maintains a packaging backlog with open tasks ranging from small single-package reviews to large multi-dependency efforts.
What you could do:
-
Package new AI/ML software for Fedora.
-
Review and co-maintain existing packages.
-
Help with version updates and build failures.
-
Test packages on different hardware (e.g., AMD, Intel, Nvidia, etc.).
-
Help migrate and organize content from the Fedora Wiki into this documentation site.
How to get started:
-
Browse the ticket tracker for tickets labeled
type/packaging. -
If you are new to Fedora packaging, start with the Fedora Package Maintainers guide.
-
Look at the current wiki content and identify pages that are outdated, incomplete, or could be better organized.
-
Submit pull requests against SIG-maintained packages and join the discussion in Matrix. Write access to the
ai-packagers-sigFAS group follows from consistent contributions, not from requesting it directly.
|
Related SIGs: Packaging work often overlaps with the PyTorch SIG, Heterogeneous Computing SIG, and ROCm Packaging SIG. In practice, these SIGs share a lot of the same members. Start by reaching out to the Fedora AI/ML SIG instead of trying to figure out which one owns what. |
Agent Skills Library
The Fedora Agent Skills Library is a collection of reusable LLM-gen-AI skills for Fedora community workflows. Agent Skills can support a wide range of AI-supported activities, including meeting support, issue triage, documentation, contributor onboarding, and other useful community workflows. The Fedora Agent Skills Library follows the Agent Skills open format. No packaging experience is required. It is open for community contributions. Ideas and feedback for new and existing Agent Skills are welcome.
What you could do:
-
Write new Agent Skills that automate or assist with Fedora contributor workflows.
-
Improve existing Agent Skills based on real-world usage and feedback.
-
Review contributions on Agent Skills from other community members.
-
Help maintain reference data and templates that Agent Skills depend on.
-
Try existing Agent Skills on relevant Fedora tasks and report bugs, gaps, or confusing instructions.
How to get started:
-
Read the CONTRIBUTING.md for how to contribute new skills, improvements, and bug fixes to the Fedora Agent Skills Library.
-
Browse the existing Agent Skills to understand the format.
-
Propose a new skill on the skills library issue tracker. See the Reviewing Agent Skills guide if you want to review contributions.
-
Submit a Pull Request when your new Agent Skill or improvement is ready.
Desktop AI Tools
Desktop AI tools are intended for Fedora developers who often work with AI tools. It aims to provide a ready-to-use environment for local AI inference, model development, and AI-assisted workflows on open source software.
This project needs contributors across many disciplines, not just packaging.
What you could do:
-
Help define what an AI-focused desktop environment should include.
-
Test builds and report issues with hardware compatibility.
-
Contribute to the container and image build pipeline.
-
Write user-facing documentation for the desktop experience.
-
Help with upstream coordination for tools like Ollama, llama-cpp, and RamaLama.
How to get started:
-
Talk to the project maintainers in the Matrix channel.
-
Follow the discussion on the Fedora Discussion tag.
User documentation
Fedora ships a growing collection of AI tools, but documentation on how to use them is scattered across wiki pages, READMEs, and upstream projects. The SIG needs help writing clear, practical guides that show Fedora users how to get started with AI on their own hardware.
No packaging experience is required. If you have used AI tools on Fedora, you have something to contribute.
What you could do:
-
Write getting-started guides for tools like Ollama, llama-cpp, RamaLama, and pi-coding-agent.
-
Document GPU setup and tuning for AMD, Intel, and Nvidia hardware.
-
Review and update existing documentation for accuracy.
How to get started:
-
Open a pull request against the ai-ml/docs repo.
-
This documentation site uses Antora and AsciiDoc. See the README for build instructions.
Other ways to help
-
Report bugs and request features: File tickets in the issue tracker.
-
Help other users: Answer questions in the Matrix channel and on Fedora Discussion.
-
Attend meetings: The SIG meets biweekly. Everyone is welcome. See the meeting info on the home page.
-
Spread the word: Write about your experience using AI tools on Fedora, share it on social media, or present at a Fedora event.
Want to help? Learn how to contribute to Fedora Docs ›