📄 Intelligent Document Processing Engine

AWS SAM Python 3.11 License: MIT

A fully serverless document processing pipeline built on AWS. Upload PDFs or images to S3 — Lambda automatically triggers AWS Textract for OCR — extracted text is stored in DynamoDB — and SNS notifications alert downstream consumers when processing completes.

Designed with event-driven architecture, IAM least-privilege policies, and infrastructure-as-code via AWS SAM.

[S3 Upload] → [Lambda: ProcessDocument] → [AWS Textract]
 ↓
[API Gateway] ← [Lambda: GetDocument] ← [DynamoDB: Results]
 ↓
 [SNS Notification]

• Event-driven S3 triggers — automatic processing on file upload
• AWS Textract OCR — extract text from PDFs, images, and scanned documents
• DynamoDB storage — document metadata and extracted text with TTL cleanup
• REST API — retrieve documents and results via API Gateway
• SNS notifications — real-time alerts on processing completion or failure
• IAM least-privilege — each Lambda gets only the permissions it needs
• Pagination support — efficiently list large document collections
• X-Ray tracing — built-in observability for all Lambda functions

• AWS CLI configured with credentials
• AWS SAM CLI
• Python 3.11+

sam build
sam deploy --guided

On first deploy, SAM will walk you through configuration (stack name, region, etc.). Subsequent deploys use the saved samconfig.toml.

aws s3 cp my-document.pdf s3://<DocumentsBucketName>/uploads/my-document.pdf

intelligent-doc-engine/
├── template.yaml # SAM infrastructure template
├── samconfig.toml # Deployment configuration
├── handlers/
│ ├── __init__.py
│ ├── process_document.py # S3 trigger → Textract
│ ├── store_results.py # Textract results → DynamoDB
│ ├── get_document.py # GET /documents/{id}
│ ├── list_documents.py # GET /documents
│ └── requirements.txt
├── tests/
│ └── test_process_document.py
├── README.md
└── LICENSE

pip install pytest boto3
pytest tests/ -v

This project is designed to stay within the AWS Free Tier for low-volume usage:

Note: Textract free tier is limited to the first 3 months. Monitor usage in the AWS Billing Dashboard.

Kushal Pitaliya

This project is licensed under the MIT License — see the LICENSE file for details.