Ask HN: How do you set up your ML Pipelines today?

What tools do you use / how do you set up back-end (model retraining/refreshing) and front-end for an ML Pipeline (say via API/microservice setup).

I've used some combination of flask/scikit-learn/docker/github/circleci/postgres, wondering if there are other setups out there that do a similar/better job.

Thanks!