Where do Google Cloud AutoML Models get deployed to?

Question

I am looking to manage all my Google Cloud machine learning models (e.g. AI Platform/Kubeflow/AutoML) in one place. However, I am unsure of where I can find the AutoML, specifically the AutoML Tables, models I have deployed.

In increasing order of complexity; I expect to be able to port the AutoML models to AI Platform models to iterate versions. Additionally, if I then want to make further customizations I could also bring that model into Kubeflow to further customize, version, manage, and deploy.

How and where are all the different models from AI Platform, AutoML, and Kubeflow managed and versioned?

Answer 1

Once you have trained a model in AutoML Tables, you will find it in Tables > Models. There, you will have the option to export it into several formats for you to deploy on different ways. From what you say, you should save it as a TF Saved Model in order to later deploy it in AI Platform. If you would also like to use Kubeflow, you can find here the corresponding Kubeflow documentation. Here there is also a Google article that talks about the several tools that it has and how to use them.