AWS Bedrock Practice Project

A hands-on project demonstrating AWS Bedrock integration with Knowledge Bases, Guardrails, and Python applications.

Project Structure

AI Project - Bedrock & Python Solution

• Streamlit Chat App (app.py) - Interactive chat interface with Bedrock LLM integration
• Bedrock Utils (bedrock_utils.py) - Core functions for model invocation, knowledge base queries, and prompt validation
• Infrastructure - Terraform modules for AWS resources:
  • Stack 1: VPC, Aurora Serverless PostgreSQL, S3 bucket
  • Stack 2: Bedrock Knowledge Base with vector storage
• Sample Data - Heavy machinery specification sheets (PDF format)

Key Features

• LLM Integration - Claude 3 Haiku/Sonnet models via Bedrock Runtime
• Knowledge Base - RAG implementation with Aurora PostgreSQL vector storage
• Prompt Validation - Content filtering for heavy machinery domain
• Infrastructure as Code - Complete Terraform deployment

Tech Stack

• Frontend: Streamlit
• Backend: Python, Boto3
• Database: Aurora Serverless PostgreSQL with pgvector
• Storage: S3 for document ingestion
• Infrastructure: Terraform, AWS Bedrock, VPC

Quick Start

1. Deploy infrastructure: terraform apply in stack1/, then stack2/
2. Run SQL setup: Execute scripts/aurora_sql.sql
3. Upload documents: python scripts/upload_s3.py
4. Launch app: streamlit run app.py

Requirements

• AWS CLI configured
• Terraform >= 0.12
• Python 3.10+
• Bedrock model access enabled