AWS Bedrock Chatbot REST API 🤖

A production-ready REST API chatbot powered by AWS Bedrock (Amazon Nova AI), DynamoDB, and FastAPI.

✨ Features

• 🌐 RESTful API - FastAPI with automatic documentation (Swagger UI)
• 🤖 Amazon Nova AI - Fast, cost-effective, no access forms required
• 💾 DynamoDB Storage - Persistent conversation history
• 🔄 Session Management - Multi-user support
• ⚙️ Easy Configuration - Change models via .env file
• 📚 Auto Documentation - Built-in Swagger UI at /docs
• 🐳 Docker Ready - Containerized deployment
• ☁️ AWS Lambda - Serverless deployment option ✅ TESTED
• ☸️ AWS EKS - Kubernetes deployment option ✅ TESTED

🚀 Quick Start (5 Minutes)

1. Install Dependencies

# Windows
install.bat

# Manual
pip install -r requirements.txt

2. Configure AWS

Option A: Create .env file (Recommended)

AWS_ACCESS_KEY_ID=your_access_key_here
AWS_SECRET_ACCESS_KEY=your_secret_key_here
AWS_REGION=us-east-1
BEDROCK_MODEL_ID=us.amazon.nova-pro-v1:0
DYNAMODB_TABLE_NAME=ChatbotConversations

Important: Use region-prefixed model IDs (e.g., us.amazon.nova-pro-v1:0 not amazon.nova-pro-v1:0)

Option B: Use AWS CLI

aws configure

3. Enable Amazon Nova in AWS Console

1. Go to: https://console.aws.amazon.com/bedrock/
2. Click "Model access" in sidebar
3. Click "Manage model access"
4. Check: ✅ Amazon Nova Pro (and optionally Lite/Micro)
5. Click "Save changes"
6. Wait 1-2 minutes (instant approval, no forms!)

4. Setup DynamoDB

python setup_dynamodb.py

5. Start the API!

python api.py

Server starts at: http://localhost:8000

6. Test It!

Open your browser: http://localhost:8000/docs

Or use cURL:

curl -X POST http://localhost:8000/chat \
  -H "Content-Type: application/json" \
  -d '{"message": "Hello! What is AWS Bedrock?"}'

📖 API Documentation

Interactive Documentation

Once the server is running:

• Swagger UI: http://localhost:8000/docs (Try it out!)
• ReDoc: http://localhost:8000/redoc (Clean docs)

Main Endpoints

Method	Endpoint	Description
GET	/health	Health check
POST	/chat	Send a message to the chatbot
GET	/history/{session_id}	Get conversation history
GET	/sessions	List active sessions
POST	/models/list	List available AWS Bedrock models

Example: Send a Message

curl -X POST http://localhost:8000/chat \
  -H "Content-Type: application/json" \
  -d '{
    "message": "Explain AWS Lambda in simple terms"
  }'

Response:

{
  "response": "AWS Lambda is a serverless compute service...",
  "session_id": "abc123-def456",
  "model": "amazon.nova-pro-v1:0",
  "region": "us-east-1",
  "timestamp": "2024-01-20T10:30:00Z"
}

🔧 Configuration

Change AI Model

Edit .env file:

# Use Nova Lite (faster, cheaper)
BEDROCK_MODEL_ID=us.amazon.nova-lite-v1:0

# Use Nova Micro (fastest, cheapest)
BEDROCK_MODEL_ID=us.amazon.nova-micro-v1:0

# Use Nova Pro (balanced, default)
BEDROCK_MODEL_ID=us.amazon.nova-pro-v1:0

Note: Amazon Nova models require region-prefixed IDs (us., eu., etc.)

Restart the API after changing.

See all available models:

curl -X POST http://localhost:8000/models/list

Override Model Per Request

curl -X POST http://localhost:8000/chat \
  -H "Content-Type: application/json" \
  -d '{
    "message": "Hello!",
    "model_id": "us.amazon.nova-micro-v1:0"
  }'

🐍 Python Client

import requests

# Send a message
response = requests.post(
    "http://localhost:8000/chat",
    json={"message": "What is AWS?"}
)

print(response.json()['response'])

# Continue conversation
response2 = requests.post(
    "http://localhost:8000/chat",
    json={
        "message": "Tell me more",
        "session_id": response.json()['session_id']
    }
)

Test Client

We provide a ready-to-use test client:

# Run demo
python test_api_client.py demo

# Interactive chat mode
python test_api_client.py interactive

# Test multiple sessions
python test_api_client.py sessions

# List available models
python test_api_client.py models

📁 Project Structure

langchain/
├── api.py                    # Main FastAPI server ⭐
├── chatbot_aws.py            # AWS Bedrock integration
├── setup_dynamodb.py         # DynamoDB setup script
├── test_api_client.py        # Python API client
├── requirements.txt          # Python dependencies
├── .env                      # Configuration (create this)
├── .gitignore               # Git ignore rules
│
├── deployment/              # Deployment files
│   ├── Dockerfile           # Docker container
│   ├── docker-compose.yml   # Docker Compose
│   ├── lambda_handler.py    # AWS Lambda handler
│   ├── lambda_deploy.py     # Lambda deployment
│   └── cloudformation_template.yaml
│
├── docs/                    # Documentation
│   ├── API_EXAMPLES.md      # Complete API guide
│   ├── ENABLE_AMAZON_NOVA.md # AWS setup guide
│   └── CHANGE_MODEL.md      # Model configuration
│
└── scripts/                 # Helper scripts
    ├── run_api.bat          # Quick start (Windows)
    └── install.bat          # Dependency installer

🐳 Deployment

Docker (Recommended for Production)

cd deployment
docker-compose up

Or manually:

docker build -f deployment/Dockerfile -t chatbot-api .
docker run -p 8000:8000 --env-file .env chatbot-api

AWS Lambda (Serverless)

cd deployment
python lambda_deploy.py

Follow the prompts to deploy to AWS Lambda + API Gateway.

💰 Costs

Amazon Nova Pricing (per 1,000 messages)

Model	Cost/1K Messages	Speed	Quality
Nova Micro	~$0.17	⚡⚡⚡	⭐⭐⭐
Nova Lite	~$0.30	⚡⚡	⭐⭐⭐⭐
Nova Pro	~$4.00	⚡	⭐⭐⭐⭐⭐

Plus:

• DynamoDB: ~$0.05 per 1,000 messages
• API calls are free when testing locally

Free Tier:

• Lambda: 1M requests/month (forever!)
• DynamoDB: 25GB storage + 25 read/write capacity units

🐛 Troubleshooting

"Cannot connect to AWS"

# Check your .env file
cat .env

# Or verify AWS credentials
aws sts get-caller-identity

"Access Denied for Bedrock"

1. Go to: https://console.aws.amazon.com/bedrock/
2. Enable Amazon Nova models (instant approval!)
3. Wait 2 minutes and restart API

"DynamoDB Table Not Found"

python setup_dynamodb.py

"Port 8000 already in use"

# Windows: Find and kill the process
netstat -ano | findstr :8000
taskkill /PID <pid> /F

# Or use a different port
uvicorn api:app --port 8001

API Returns Errors

1. Check logs in the terminal running api.py
2. Verify AWS credentials in .env
3. Ensure DynamoDB table exists
4. Confirm Bedrock models are enabled

📚 Documentation

Getting Started

• Quick Start: START_HERE.md - New to the project? Start here!
• API Examples: docs/API_EXAMPLES.md
• Model Configuration: docs/CHANGE_MODEL.md
• Model IDs Reference: docs/MODEL_IDS.md ⭐ (Important!)
• AWS Setup: docs/ENABLE_AMAZON_NOVA.md

Deployment

• Deployment Overview: docs/DEPLOYMENT_SUMMARY.md - Quick reference
• Deployment Guide: deployment/README.md - Choose your method
• AWS Lambda: deployment/DEPLOY_LAMBDA.md - Serverless
• Kubernetes (EKS): deployment/DEPLOY_KUBERNETES.md - Production
• Architecture: deployment/ARCHITECTURE.md - System design

Testing & Reference

• Testing Guide: docs/TESTING_GUIDE.md - Test local & AWS deployments
• Project Structure: docs/PROJECT_STRUCTURE.md - File organization
• Cleanup Summary: docs/CLEANUP_SUMMARY.md - What was changed

🛠️ Development

Running Locally

# Start API
python api.py

# In another terminal - test
python test_api_client.py interactive

Environment Variables

Create a .env file:

AWS_ACCESS_KEY_ID=xxx
AWS_SECRET_ACCESS_KEY=xxx
AWS_REGION=us-east-1
BEDROCK_MODEL_ID=amazon.nova-pro-v1:0
DYNAMODB_TABLE_NAME=ChatbotConversations

Requirements

• Python 3.8+
• AWS Account with Bedrock access
• AWS credentials configured

🌟 Why Amazon Nova?

✅ No Access Forms - Instant approval
✅ Lower Cost - Up to 4x cheaper than alternatives
✅ Fast Performance - Quick responses
✅ Multimodal - Nova Pro supports images
✅ Built by AWS - Native integration

🔗 Links

• AWS Bedrock: https://aws.amazon.com/bedrock/
• FastAPI Docs: https://fastapi.tiangolo.com/
• LangChain: https://python.langchain.com/

📝 License

MIT License - Feel free to use and modify!

🤝 Contributing

Suggestions and improvements welcome! Key areas:

• Streaming responses
• Authentication/Authorization
• Rate limiting
• Response caching
• Multi-modal support (images)
• Conversation summarization

---

Built with AWS ☁️ | Powered by Amazon Nova 🤖 | Production-Ready 🚀

Quick Start: install.bat → Configure .env → python api.py → Visit http://localhost:8000/docs