Model Versioning

[jsmith]

This isn't a super important issue now, but I really want to look into model versioning.

Once we finally deploy models to production, how do we update them? Ideally, the update process should be independent of the application deployment process. Maybe a user receives a notification that a new version is available and gives them the option to download the updated version? Do we just deploy new models as a service?

The scope of this issue doesn't concern the whole update process. Eventually, we will have to create another issue in the DAW repository.

See:
https://blog.algorithmia.com/how-to-version-control-your-production-machine-learning-models/