Ideally, a user in a CI/CD environment would like to:

• deploy a given cluster
• deploy a given plugin or set of plugins
• deploy a specific set of artifacts (it's training code)
• get the outcome and save it
  This should be done through a simple config file (YAML?) that allows the above steps to be represented and embed the training in the form of existing defaults (i.e.: KFP, MLFlow, etc..) or as a single file (.py)