Getting Started with Amazon SageMaker

Introduction & Lesson Overview

Welcome to the next step in your machine learning journey—moving from training models locally to building and deploying them in the cloud with Amazon SageMaker . In the previous course, you learned how to build machine learning workflows on your local machine, including data preprocessing, model training, evaluation, and deploying models as REST APIs. Now, we are taking the next step by moving these workflows to the cloud using Amazon SageMaker . This transition will allow you to leverage the power and flexibility of cloud computing, making it easier to scale your machine learning projects and collaborate with others. In this lesson, you will get an overview of what Amazon SageMaker is, what it can do for you, and how to set up your environment so you are ready to start building and deploying models in the cloud.

What Is Amazon SageMaker?

Amazon SageMaker is a fully managed service from AWS that helps you build, train, and deploy machine learning models at scale. Instead of managing your own servers or worrying about infrastructure, SageMaker provides everything you need in one place. You can use it to prepare your data, train models using powerful cloud hardware, deploy models as APIs, and monitor their performance — all from a single platform. SageMaker is designed to support the entire machine learning lifecycle, making it easier for you to go from an idea to a production-ready model.

Core Capabilities Of SageMaker

SageMaker offers several key features that cover every stage of the machine learning process. For data preparation, you can use built-in tools to clean and transform your data or connect to data stored in AWS services like S3 . When it comes to model training, SageMaker lets you choose from popular built-in algorithms, bring your own code, or use pre-built containers for frameworks like Scikit-Learn, TensorFlow and PyTorch . Once your model is trained, SageMaker makes it easy to deploy it as a REST API endpoint so you can make predictions in real time. It also provides tools for monitoring your deployed models, tracking their performance, and detecting issues like data drift. For example, you might use SageMaker to train a model on a large dataset stored in S3, deploy it as an endpoint, and then monitor its accuracy as new data comes in.

Join the 1M+ learners on CodeSignal

Be a part of our community of 1M+ users who develop and demonstrate their skills on CodeSignal

What Permissions Would You Need?

While you don't need to set up any permissions in this course (everything is already configured for you), it's helpful to understand what would be required if you were setting up SageMaker in your own AWS account. This knowledge will help you understand error messages and work effectively with cloud infrastructure teams in the future.

To use SageMaker effectively, you would need two distinct sets of permissions:

1. Permissions for Your User Account

Your IAM user needs permissions to interact with SageMaker and related services. This includes:

Creating and managing SageMaker resources (training jobs, endpoints, etc.)
Uploading data to and downloading results from S3 (AWS's storage service)
Granting SageMaker permission to act on your behalf (the iam:PassRole permission)

AWS provides a managed policy called AmazonSageMakerFullAccess that includes these permissions. While convenient for learning and development, production environments typically use more restrictive custom policies.

2. Permissions for SageMaker (The Execution Role)

When SageMaker performs tasks for you, it needs its own permissions through an IAM role. This execution role must be able to:

Access your training data in S3
Write trained models and other outputs back to S3
Send logs to CloudWatch for monitoring and debugging
Pull machine learning framework containers from ECR (AWS's container registry)

You'll often see this role referenced in SageMaker operations as an ARN (Amazon Resource Name) that looks like: arn:aws:iam::123456789012:role/SageMakerExecutionRole. In this course, both sets of permissions are pre-configured in your environment. However, if you work with SageMaker outside this course and encounter errors like "AccessDenied" or "not authorized to perform: iam:PassRole," it typically indicates that one of these permission sets needs adjustment.

The key point: AWS enforces strict permission boundaries to keep your resources secure. While this adds some initial complexity, it ensures that your data and models remain protected in the cloud.

Introduction & Lesson Overview

What Is Amazon SageMaker?

Core Capabilities Of SageMaker

Join the 1M+ learners on CodeSignal

How Amazon SageMaker Can Help

Get Your AWS Credentials in Place

Understanding AWS Permissions (IAM Basics)

What Permissions Would You Need?

Installing Required Python Packages

Lesson Summary & Next Steps