Opening an AWS bill after experimenting with SageMaker led to a $956 charge, emphasizing the need to manage machine learning costs effectively.
Lessons learned include the importance of stopping notebook instances when not in use, starting with smaller instances for training to save costs, deleting test endpoints after usage, and setting up S3 lifecycle rules for data management.
Cost-saving strategies include setting billing alarms, utilizing Cost Explorer to track spending, adding tags to projects for cost tracking, following a shutdown ritual to manage resources effectively, and never signing out of AWS without checking for running instances.
By following these guidelines and best practices, individuals can avoid unexpected costs and build innovative projects on AWS without financial stress.