In this particular genre there are no off-the-shelf platforms or even accepted set of best practices for something as comprehensive as this for generalized ML p

In this particular genre there are no off-the-shelf platforms or even accepted set of best practices for something as comprehensive as this for generalized ML pipelining infrastructure. It's still very Wild West. I've done a lot of research into what pipelines look like at different companies and cloud providers and I can tell you this is a nice one. Some have a good training story, some have a good deployment story, some have a good experimentation story, and so on but none have all the things. Projects like Tensorflow Serving and PipelineAI are heading there but still quite a lot missing compared with something like what was presented here.