AWS Builder Center
CROSSCHECK: Detecting Public Infrastructure Collisions with Bedrock, Lambda, and DynamoDB

CROSSCHECK: Detecting Public Infrastructure Collisions with Bedrock, Lambda, and DynamoDB

How we built a serverless pipeline on AWS that reads tender PDFs with Amazon Bedrock and automatically flags road-work conflicts before contractors dig in the wrong place at the wrong time.

AI & ML Enthusiast | Building AI-Powered Solutions | CSE Student
Public infrastructure projects often exist in isolation.
One agency publishes a tender to repair a road. Another plans drainage work on the same stretch. A third schedules utility work a few weeks later. The information exists, but it is scattered across documents, making it difficult to spot these overlaps early.
That’s the problem we wanted to tackle with CROSSCHECK — Public Infrastructure Collision Detector.
Our pipeline starts with something very simple: a tender PDF.
“Here’s how a tender gets in. I upload the PDF straight to S3 with a presigned URL. That triggers a Lambda, which calls Amazon Bedrock to read the document — no templates, no regex. Bedrock pulls out the agency, the road segments, the start and end dates, the contract value. That lands in DynamoDB, and the collision engine re-runs against every other tender in the table.”
Once the tender is structured, CROSSCHECK compares it against existing projects using spatial and temporal overlap. If two projects affect the same or nearby road segments within overlapping or closely spaced time windows, the system flags the potential collision.
The result is a dashboard where infrastructure teams can see where projects overlap, which agencies are involved, the relevant timelines, source evidence, and the estimated expenditure at risk.

Building it on AWS

The entire application runs on AWS, deployed in ap-south-1.
  • Amazon S3 — hosts the frontend and stores raw tender PDFs
  • Presigned URLs — allow direct tender uploads to S3
  • AWS Lambda — runs the document-processing pipeline and Python collision engine
  • Amazon Bedrock — extracts structured information from unstructured tender documents
  • Amazon DynamoDB — stores tenders and detected conflicts
  • Amazon API Gateway — exposes the backend APIs to the frontend
  • AWS CDK / CloudFormation — defines the infrastructure as code
“The entire stack is one CloudFormation deployment — tear it down, redeploy it, same infrastructure.”
For us, the interesting part wasn't just putting an AI model behind a UI. It was connecting document intelligence + cloud infrastructure + geospatial reasoning into one pipeline that addresses a real-world coordination problem.
That’s CROSSCHECK: turning scattered public infrastructure documents into something that can actually be compared, mapped, and acted on.

Any opinions in this article are those of the individual author and may not reflect the opinions of AWS.
Enjoyed reading this content? Let the author know!

Your likes, comments, shares, and saves help creators reach more builders.

Loading recommendations

Loading article