Anyone know of a beginning to end Ml platform I might be looking for? I am on the sw eng side, not data science, so I can create novel / hello-world models in python and serve them as rest endpoints with flask, or MLFlow fine …

but like all things corporate we need “project managers” and “data scientists” to be able to take a spreadsheet, figure out 16 features, categorize based on that (regression, etc), make a model with that … and then send it off to me to serve it. Our endpoint needs to take a record in, predict, and return with subsecond replies. Totally doable for our stuff.

For some reason they clamped onto Alteryx. Fine. Except getting models out into sklearn/pickle formats I can load into MLFlow is not something “they” will want to do. Alteryx won’t run as a rest capable endpoint either.

automl stuff is cool too.

Is there a toolchain that can let them experiment, tinker, tune and … possibly send to my flow server’s logging setup for release later?