sagemaker-simple-demo

Simple demo of using a Sagemaker Studio notebook to train a model and deploy an API endpoint. This is based on the great AWS Sagemaker Jumpstart notebook but uses the iris dataset instead of mnist. I also removed some steps and made the code less sophisticated in a couple places to make it as minimal of an example as possible. If you are trying to learn Sagemaker, this notebook demonstrates a few critical pieces that are specific to Sagemaker:

1. The usage of a Sagemaker "Session" and a Sagemaker execution role
2. Converting training data to recordIO-wrapped protobuf format
3. Defining an input S3 data location and an output S3 location
4. Deploying an endpoint

These steps can be a bit confusing when you're just starting, but they do help Sagemaker be tightly integrated with other AWS services and also help with scalability.