Machine Learning with Amazon Sagemaker studio

Amazon Sage maker is a fully web based machine learning development environment. It provides the tools to build, train and deploy ML models in one platform using AWS. You can manage all steps from preparation to deployment using this tool.

Understanding SageMaker tools

A sagemaker domain is an environment where SageMaker studio and other tools operate. Some of the key components are -

1. Integration with VPC’s: SageMaker Domains can connect to Customer VPCs (default or internal), allowing access to other instances via Elastic Network Interfaces.

2. EFS Store: Every SageMaker Domain includes an EFS (Elastic File System) store for storing files like notebooks, datasets, and other resources.

3. IAM Execution Roles: SageMaker Domains require an IAM Execution Role to control access to AWS resources like S3 buckets and ensure secure operation within your AWS account.

Pricing considerations

First thing is to consider is domain and studio setup. There is no monthly cost for domain or studio, aside from minimal storage costs of S3 or EFS.

However Notebooks and Instances costs because of compute resources. Training jobs and inference endpoints also incur costs depending on the instance type.

Steps to set up Sagemaker studio

1. Access the Sagemaker console by login/signup to the aws management console.

2. Create an IAM role and select data scientist persona for permissions, also add SageMakerFullAccess

   and S3FullAccess policies

3. Configure a SageMaker domain and leave it to the default settings(unless required)

4. Launch the SageMaker studio

Exploring SageMaker Studio

Key Features of the Interface

1. Homepage:
   The SageMaker Studio homepage provides access to tools like Jumpstart, which offers pre-built models and workflows.

2. Sidebar Tools:
   The left-hand sidebar offers navigation to tools like notebooks, datasets, and file storage.

3. File Browser:
   View and manage files stored in your EFS volume directly within Studio.

Shutting Down Resources

To avoid unnecessary costs:

1. Stop the Kernel:

   • In the sidebar, locate the running instance and shut it down.

   • This stops the EC2 instance while preserving your notebook and its outputs.

2. Close the Notebook:

   • Close the tab and choose not to save changes unless needed.

Conclusion

In this guide, we explored how SageMaker Studio integrates into the ML lifecycle, the role of SageMaker Domains, and the basics of setting up and navigating the Studio interface. With these fundamentals in place, you're ready to start developing and managing ML models in a powerful, fully integrated environment.