What is the use of an AWS SageMaker for a data scientist of an ML engineer?

Two primary uses that I have experienced:

  1. Quick iterations over different types of models. This is mostly applicable to new ML engineers who are not really sure of which model would best suit their problem-in-hand. They can very quickly prototype few models from the templates already available in Sagemaker

  2. Scaling of the model after go-live. Most of the times, the model gets trained and tested on a subset of actual data so the performance implications are not really evaluated by the engineer. But, with live data most models run into scalability and performance issues. Sagemaker solves this problem quite well by almost seamless scalability