HydroSentry: Building an Urban Flood Sentinel with AWS Open Data, Cedar, and Serverless

HydroSentry is a real-time urban flood monitoring and automated emergency dispatch system built for vulnerable road underpasses in Delhi NCT, India.

Sreenandh S

Published Oct 11, 2026

Last modified Oct 11, 2026

Why We Built HydroSentryDuring every Delhi monsoon, low-lying underpasses like Minto Bridge and Pul Prahladpur fill with water in less than 30 minutes. Navigation apps only report traffic slowdowns after vehicles are already stuck, and municipal pump teams react hours late. We built HydroSentry for the AWS Environmental Hacks Hackathon to predict underpass flooding 20 to 30 minutes before water actually pools.
The Stack & ArchitectureInstead of using complex machine learning models, we used physics: rainfall rate minus drain capacity.
AWS Open Data: We used 30-meter elevation data from the Copernicus Digital Elevation Model (hosted on AWS S3) to map natural depression bowls.
AWS Serverless: Built using AWS SAM, AWS Lambda, API Gateway, and Amazon DynamoDB with Time-To-Live (TTL) so expired flood warnings clear automatically.
AWS Cedar: Used AWS Cedar policy language to strictly authorize emergency pump dispatches only when water depth exceeds 350 mm and failure is imminent.
What Fought BackOur biggest hurdle was avoiding false alarms. If rain is heavy for only 5 minutes, normal drains handle it. We had to ensure Cedar checked both water depth and time-to-submersion so civic teams aren't sent out unnecessarily.
LinksGitHub Repository:  https://github.com/Naman26331/HydroSentry 

# serverless # open-source # python

Builder Center

HydroSentry: Building an Urban Flood Sentinel with AWS Open Data, Cedar, and Serverless

Sreenandh S

Why We Built HydroSentry

The Stack & Architecture

What Fought Back

Links