2.0 KiB
2.0 KiB
SIG/AI - 2023-04-06
Rollcall
* Balaji B
* Mark Watson
* Ivan Novick
* Rajkumar Rajendran
* Neil Hanlon
* Sherif Nagy
* Gregory Kurtzer
* Arthur Tyde
* Forrest Burt
* Alexandre Barbieri
Links
* https://sig-ai.rocky.page/
Purpose
To showcase and deliver user-friendly, robust, scalable AI ecosystems.
Goals
* deliver end-to-end open-source enterprise OS with flexibility to setup as per user needs
* bring in popular FOSS AI frameworks/libraries as technical partners of RockyAI
* become top OS among AI user community and be part of AI applications being developed and deloyed
* rank top in AI ecosystem deployed across BM, Cloud, Containers
Objectives
* Publish docs explaining installation and usage of popular AI frameworks, job schedulers, cluster managers, and AIOps
* Work with maintainers of popular frameworks to include in Rocky
* Publish cloud and container images in collaboration with partners for easy and scalable deployment
* Release RockyAI toolset for monitoring and maintenance of AI stack
* Hosting virtual conference demonstrating RockyAI
Roadmap
| Time | Tasks |
|---|---|
| April - June | Publish RockyAI Docs and Find partners to work and collaborate with. Begin developing RockyAI Toolset |
| July - Aug | Release RockyAI Toolset |
| Sep - Oct | Publish prepackaged partner images |
| Nov - Dec | Continue packaging and automating publishing of images |
| January 2024 | Virtual user conference on RockyAI & other SIG's |
| Feb - March 2024 | Expanding the RockyAI partnership with more applications |
New business / Discussion
* Arthur - some packages we can control, some we cannot, or cannot ship them due to being nonfree, etc.
* EPEL is a good option, want it to be simple
* Collaborating with NeuroFedora is a very good idea
* https://docs.fedoraproject.org/en-US/neurofedora/ -
* Want to be able to build and release packages
* Longer term, want to be able to add cloud images, etc
Packages to target initially
* python frameworks
* tensorflow
* pytorch
* apache spark