Foundation Model Integration, Data Management, and Compliance II
Explore how to implement effective data validation, preprocessing, and prompt governance strategies for generative AI systems on AWS. Understand techniques for managing document quality, transforming diverse data types, and enforcing audit-compliant prompt controls to optimize foundation model integration and retrieval-augmented generation workflows.
We'll cover the following...
Question 6
A company is building a document processing pipeline that ingests PDF files from Amazon S3 before sending extracted text to an Amazon Bedrock foundation model for summarization. The company wants to ensure data quality standards such as non-empty content, minimum text length, and valid metadata before the documents are consumed by the foundation model.
Which solution is the most appropriate for implementing these validation checks?
A. Use AWS Glue Data Quality rules to validate extracted text and metadata before passing the data to the FM.
B. Invoke the Amazon Bedrock model and rely on prompt instructions to reject invalid documents.
C. Use Amazon Comprehend to detect entities and infer whether ...