This lab provides hands-on training in building a secure generative AI chatbot using Amazon Bedrock. Participants will learn how to leverage Retrieval-Augmented Generation (RAG) to generate contextually relevant responses. They will implement guardrails to filter and control the chatbot's content generation. Additionally, participants will explore essential security features like access control and logging to ensure the Bedrock configuration adheres to security best practices. By completing this lab, participants will acquire skills to develop secure and compliant generative AI applications aligning with organizational requirements and ethical guidelines.
Objectives
- Enable model access in Amazon Bedrock.
- Set up a knowledge base using provided sample files in an S3 bucket.
- Test the knowledge base with sample prompts.
- Enable prompt attack filters and handle Personally Identifiable Information (PII) data.
Prerequisites
- Basic knowledge of Generative AI and RAG architecture.
- Basic knowledge of AWS services.
Outline
Task 1: Enable model access for Titan model
Task 2: Setup the Knowledge Base
Task 3: Test the knowledge base with sample prompts
Task 4: Set Up Guardrails
Task 5: Testing Guardrails
