Intro to AWS SageMaker for Predictive ML/AI
About this resource
This introductory AWS SageMaker workshop teaches core workflows for running predictive ML/AI models in AWS SageMaker, an AWS-managed machine learning environment. Participants will learn to set up data, configure SageMaker Notebooks, manage code repositories, train and tune models, and optimize resource costs effectively within AWS. Users will benefit from tips on controlling AWS expenses and scaling models efficiently, with real-world guidance on choosing appropriate CPU and GPU resources.
Cost estimate
Running through this workshop should cost approximately $5-$10 on AWS, assuming moderate usage of GPU instances and a few parallel jobs (i.e., sticking to the lesson materials). For new AWS accounts, the AWS Free Tier may cover some of these costs, including 250 hours per month of the ml.t2.medium instance for the first two months, as well as some limited S3 storage. New users may be able to complete certain parts of the workshop for free or at a significantly reduced cost. We recommend monitoring usage through the AWS Billing Dashboard to stay within the free tier and manage any extra expenses effectively.
Prerequisites
Estimated time to complete
3-5 hours: Based on running through training, tuning, and experimenting with example code setups.
Questions?
For any questions, please post to the Nexus Q&A on GitHub. Feedback is especially helpful in these early stages to improve workshop materials!
See also
- AWS Free Tier Guide: An overview of the AWS Free Tier, including limitations and expected costs for beginner users.
- Compute: BadgerCompute – UW–Madison’s lightweight, NetID-authenticated Jupyter service for short interactive sessions and classroom use. Includes a 4-hour runtime limit (which may sometimes beat the free version of Colab).
- Compute: Google Colab - Learn how to use Google Colab for machine learning workflows.
- Compute: Center for High Throughput Computing (CHTC) - Learn how to use CHTC for machine learning jobs.