[No Action, Just Support!]: From the Co-founder of Kubeflow #484
Description
Hey AICR team!
I wanted to drop a note here because this project genuinely made my day.
I was very very early on Kubernetes at Google, and then co-founded Kubeflow. The whole dream from day one was simple: you show up with your infrastructure (a VM, a disk, a Kubernetes cluster, whatever) and we give you a declarative way to roll out your ML stack on top of it. No more bespoke scripts, no more tribal knowledge locked in some engineer's head, no more "well it works on my cluster."
We got in the neighborhood with Kubeflow, but honestly, we never fully got there. The gap between "here are my resources" and "I'm running ML workloads" still required way too much glue, too much context that lived in people's heads or was buried in Terraform scripts that nobody else could read.
AICR makes HUGE strides here. The layered metadata approach, where you have a baseline and then apply cloud-specific or accelerator-specific overrides that merge cleanly, is exactly the kind of architecture that makes declarative infrastructure rollout actually work across the real-world diversity of environments. This is what I was dreaming about when we started Kubeflow, and you've taken it further than I thought anyone would.
I'm genuinely excited about where this goes. If there's any way I can help, whether that's bridging to the Kubeflow community, making introductions to research labs and organizations I know in the cloud-native ML space, or just being a loud voice of support, I'm here for it. Don't hesitate to reach out.
— David Aronchick (co-founder, Kubeflow)
aronchick (at) gmail (dot) com
Are you willing to contribute?
Yes, I can open a PR