This article demonstrates how to integrate AWS Lake Formation with Amazon SageMaker Data Wrangler and Amazon EMR to enable fine-grained data access controls for ML data preparation.

• SageMaker Data Wrangler now supports Lake Formation with Amazon EMR for data access control
• Lake Formation provides centralized governance with fine-grained permissions across accounts
• Solution uses two AWS accounts: data lake account (A) and data science account (B)
• CloudFormation template automates deployment of EMR, IAM roles, SageMaker Studio, and Lake Formation
• Different users see different tables based on Lake Formation permissions (e.g., David sees non-sensitive customer data only)
• Data Wrangler connects to EMR via Hive/Presto using IAM authentication
• Users can transform data visually without writing code, then export to SageMaker Feature Store
• EMR encryption in transit required; self-signed certificates for proof-of-concept only

This solution enables data scientists to prepare ML data securely with granular access controls while maintaining a no-code visual interface.