Cloud Native playground for AI workloads

[rajaskakodkar]

This is an initiative under WG AI for a playground where contributors can:

• deploy AI workloads
• try out different reference architectures
• tinker with MLOps pipelines
• identify and explore the gaps in deploying AI workloads in cloud native infrastructure
• understand how projects like Kubeflow allow you to do the full ML lifecycle: data prep, training, creating/hosting AI/ML artifacts, serving and observability (in a small environment)
• provide leadership boards for reference architectures including other details like carbon footprint of hosting AI workloads

This is an umbrella issue and requires a bunch of aspects which need to be considered (the following list is work in progress)

• Identifying the platform to host this playground
• Designing this playground
• Identifying cloud credits for the playground

[rajaskakodkar]

cc @raravena80 @cathyhongzhang @nikhita

[mkbhanda]

Started https://docs.google.com/document/d/1oIebBZy06SulzQvXykJ7-151n1zrHc4hFo7uormhJuk/edit?userstoinvite=nikitaraghunath%40gmail.com&sharingaction=manageaccess&role=writer to collect input from interested parties