Hello all, I'm Riley and I'm an ML Platform Engineer. I am evaluating tooling for facilitating model training, experimentation, and workflow orchestration as part of our initiative to introduce a unified ML Platform used across our entire organization. About 4 months ago, I created this post on the Machine Learning subreddit and I received several comments from individuals recommending Metaflow. We are boxed into using AWS and Sagemaker, so Metaflow would make a great fit given that it can leverage existing AWS tools like step functions and batch. Was hoping to do a POC at some point and will have some questions along the way. Also wondering if Metaflow is completely open source. Is there a paid enterprise version?

Hi Riley 👋 Metaflow is and will be completely open source If you want a managed platform for running Metaflow flows, Outerbounds can help 🤗

Thank you very much for sharing this information @straight-shampoo-11124, I think at this point we are primarily interested in evaluating open source offerings and would like to either deploy multiple instances of this infrastructure either in our stakeholders' accounts or have a centralized instance in own designated Machine Learning Platform AWS account. That is still to be determined. We haven't yet nailed which approach would work better - multi-tenancy is probably one of the biggest hurdles that we're facing when evaluating these orchestration tools. Metaflow looks promising though. Was thinking of leveraging that production-grade terraform bundle and then using our in-house spinnaker to deploy into our customer accounts. That pattern might work well...Would definitely appreciate your insights on this.

interesting. If you want, we could have a quick zoom call so I can share more. Find a time that works for you here

That would be great! Thanks @straight-shampoo-11124