The problem
Airport licensing approval at the Airport Authority of India meant officials manually reading official documents to pull out the tables and key facts that drive the decision. Slow, repetitive, and error-prone: exactly the kind of bottleneck automation is for.
The approach
We built a web application for the approval workflow, backed by blockchain for tamper-resistant records. The document intelligence runs serverless: when a document lands in an S3 bucket, a Lambda function fires, uses AWS Textract to extract tables and key-value pairs, and pushes the structured information straight into the app's frontend for the approver.
The outcome
Selected for the Smart India Hackathon 2020 Grand Finale, organized by MHRD, Government of India, and finished Runner-up among the top 5 finalist teams.
Why it matters
This was my first taste of the full product loop: find a real institutional pain point, ship something working under a deadline, and defend it in front of judges. It's also the project that made document intelligence and serverless architecture feel like practical tools rather than buzzwords.