Defining a self-service platform

Self-service is defined as the capability of a platform that allows platform end users to provision resources on-demand without other human intervention. Take, for example, a data scientist user who needs an instance of a Jupyter notebook server, running on a host container with eight CPUs, to perform his/her work. A self-service ML platform should allow the data scientist to provision, through an end user friendly interface, the container that will run an instance of the Jupyter notebook server on it. Another example of self-service provisioning would be a data engineer requesting a new instance of an Apache Spark cluster to be provisioned to run his/her data pipelines. The last example is a data scientist who wants to package and deploy their ML model as a REST service so that the application can use the model.

One benefit of a self-service platform is that it allows cross-functional teams to work together with minimal dependencies on other teams...