[KR2] Data Center Demo

[MichaelClifford]

Choose a data set and use case for a model training workflow in the datacenter cluster. Run this workflow using the Ray stack (Ray data, Ray tune, Ray horovod, etc)

• [x] Find a dataset for a reasonably interesting use-case
• [x] Prepare dataset
• [x] Develop training pipeline
• [x] Leverage GPU's in training pipeline
• [ ] Record demo of workflow
• [ ] Develop model monitoring capabilities* (stretch to support KR5)

Acceptance Criteria:

• [ ] This issue is done when we have Demo running on the data center cluster that uses as much of the Ray stack as possible