Support for Machine Learning and AI

[jonmjoyce]

What are some common workflows or capabilities that should be considered to make the Next-Gen DMAC ready for a future supporting ML/AI applications?

A few starter ideas:

1. Fast, scalable access to datasets across large dimensions (e.g. years worth of data over time)
2. The ability to add and manage labelled metadata tags to datasets for training purposes