π About the Project
π Inspiration
The motivation behind this project came from the growing need for quick content understanding in the age of information overload. Whether it's lengthy documents, scanned images people need instant summaries without reading every line. This project aims to solve that problem by offering an AI-powered document summarization tool accessible via a simple web dashboard.
π οΈ How I Built It
Frontend: Built a responsive dashboard using HTML and Bootstrap, allowing users to upload
.txt,.jpg, files or enter text directly.Backend: Implemented using AWS Lambda, API Gateway, and Step Functions to handle file processing and summary generation.
Storage:
- Inputs and outputs are stored in separate Amazon S3 buckets (
doc-summary-inputs-project&doc-summary-outputs-nivas). - DynamoDB is used to store metadata such as
summary_id,user_id, and file types.
- Inputs and outputs are stored in separate Amazon S3 buckets (
Processing:
- For
.txtfiles, the text is extracted directly and summarized. - For images the file is uploaded to S3, and downstream services like Amazon Textract (if added) can extract text for summarization.
- For
Security: IAM roles and bucket policies were configured carefully to enable access for Lambda, API Gateway, and EventBridge.
π§ What I Learned
- How to integrate multiple AWS services into a single serverless architecture.
- Handling CORS issues between S3 static websites and API Gateway.
- Using Step Functions for orchestrating multi-step workflows.
- Managing file uploads (binary to Base64) and extracting text in different formats.
π§ Challenges Faced
- CORS Issues: Had to explicitly set CORS headers on API Gateway and Lambda responses for frontend access.
- File Format Handling: Supporting different file types (text, image) required careful encoding and decoding logic.
- Dynamic Folder Structure: Ensuring user-specific folders in S3 to organize uploads without collisions.
- Latency: Summary generation could take 15β30 seconds. We handled this with async processing and status checks.
β Whatβs Next
- Add real-time notifications using Amazon SNS or WebSockets.
- Enable multi-language support for global users.
Built With
- amazon-dynamodb
- amazon-web-services
- apigateway
- bedrock
- boto3
- lambda
- python
- s3
- stepfunctions
- textract

