Setting up Amazon SageMaker Environment On Your Local Machine

Photo by Erico Marcelino on Unsplash

Amazon SageMaker is beyond just managed Jupyter notebooks, it is a fully managed service that enables you to build, train, optimize and deploy machine learning models. A common misconception, specially when you are starting out with SageMaker is that, in order to use these services, you need a SageMaker Notebook Instance or SageMaker (Studio) Notebook. You can in fact kick off all these services directly from your local machine or even from your favorite IDE.

Before we go further, let’s consider how we interact with Amazon SageMaker services. You have two APIs

SageMaker Python SDK — This is a high level API in Python that abstracts the code to build, train and deploy machine learning models. Specifically it provides estimators for first class or built in algorithms as well as supported frameworks like TensorFlow, MXNET etc. In most cases you will use this to interact with your interactive machine learning tasks.

AWS SDK — This is a low level API that is used to interact with all supported AWS services, not specific to SageMaker. The AWS SDK is available in most popular programming languages like Java, Javascript, Python (boto) etc. In most cases you will use this service-level APIs for things such creating resources for automations or interacting with other AWS services that are not…