Implementation Week 2 · Thursday

Day 11: Hands-On Build - RAG App with Guardrails (Part 1)

Learning Objectives

0/3 completed

Read30m
Bedrock Knowledge Base Tutorial
Step-by-step guide to creating a Knowledge Base in the console.
Blog25m
Guardrails Content Filters, PII, and Streaming
Follow this as a practical guide for setting up guardrails.
Hands-on120m
Build: RAG App with Guardrails (Steps 1-5)
1) Create S3 bucket + upload PDF, 2) Create KB with OpenSearch + Titan Embeddings, 3) Configure chunking, 4) Test retrieval, 5) Create Guardrail with content filters, PII, contextual grounding.

Write your understanding, then reveal the reference answer.

0/3 reviewed

Build real muscle memory with these activities.

intermediate 60 min

Build the first half of the RAG app by creating the KB and testing chunking options.

1 Create an S3 bucket called 'rag-build-data' and upload 3-5 PDF documents
2 Create a Bedrock Knowledge Base with OpenSearch Serverless and Titan Embeddings v2
3 Set chunking to 'Fixed size' with 300 tokens and 20% overlap
4 Sync the data source and test retrieval with 3 queries in the console
5 Note the chunk relevance scores for comparison later

intermediate 45 min

Configure a comprehensive Bedrock Guardrail with multiple filter types.

1 Open Bedrock → Guardrails → Create guardrail
2 Add content filters: set HATE, INSULTS, SEXUAL, VIOLENCE, MISCONDUCT to HIGH strength
3 Add PII detection: enable SSN, credit card, email with ANONYMIZE action
4 Add a denied topic: 'competitor product comparisons'
5 Enable contextual grounding with threshold 0.7 for grounding and 0.5 for relevance

18 questions across 3 difficulty levels.