AWS Machine Learning Best Practices Report

Overview

This report assesses AWS resources against 16 enabled configuration checks across the areas below. It identifies passed and failed checks, unavailable evidence, and affected resources so you can prioritize configuration changes and investigate operational gaps.

SageMaker Notebook Instances

• SageMaker Endpoint Config Production Redundant Instance Check
• No Direct Internet Access for SageMaker Notebook Instance
• SageMaker Notebook Instances Launched in a Non-Default VPC
• SageMaker Notebook Instance Root Access Disabled
• SageMaker Notebook Instance Encryption At-Rest Enabled
• SageMaker Notebook Instance in VPC
• SageMaker Notebook Instances Run On Supported Platforms

SageMaker Endpoints

• SageMaker Endpoint Configuration Encrypted At-Rest

SageMaker Training Jobs

• SageMaker Training Job Inter-Container Encryption Enabled
• SageMaker Training Job Network Isolation Enabled
• SageMaker Training Job Volume Encryption Enabled with KMS
• SageMaker Training Job has VPC Settings Enabled

SageMaker Models

• SageMaker Model Network Isolation Enabled
• SageMaker Model VPC Settings Enabled
• SageMaker models should use private registry in VPC for multi-container inference pipelines

Glue ML Transforms

• Glue ML Transform Encryption at Rest Check

The report includes actionable findings and remediation guidance for supported checks.