Mathura and Vrindavan attract large numbers of pilgrims, especially during major religious occasions. Navigating the region can be difficult because visitors need information about temples, routes, timings, parking and local guidance, while authorities face a different challenge: understanding changing crowd conditions and identifying potential risk before a situation becomes critical.
We wanted to build a single platform that could help both sides — pilgrims looking for reliable guidance and authorities monitoring crowd conditions.
What We Built
We built BrajSahayak, an AI-powered pilgrimage and crowd-safety decision-support platform.
For pilgrims, the platform provides a conversational interface for asking questions about their visit and accessing useful local information.
For authorities, BrajSahayak provides a crowd-monitoring dashboard where crowd conditions can be simulated and monitored using factors such as occupancy, capacity, inflow and outflow. A deterministic risk engine identifies changing risk levels, while the AI layer converts the structured information into an understandable situation brief.
Our core workflow is:
Detect → Explain → Help Humans Decide
The goal isn't to replace authorities, but to provide them with clearer information to support human decision-making.
AWS Stack
We deployed the application using AWS serverless services:
AWS Amplify — frontend hosting and deployment
Amazon API Gateway — API communication between frontend and backend
AWS Lambda — serverless backend logic and processing
Amazon DynamoDB — application data
Amazon S3 — cloud storage
This architecture allowed us to build and deploy the application without managing traditional servers.
What Fought Back
The biggest challenge was not simply writing the application — it was understanding how all the cloud components fit together.
As beginners to AWS, connecting the frontend with API Gateway and Lambda, configuring deployments, managing environment variables, handling permissions, and debugging cloud-side issues required a lot of experimentation.
Another challenge was deciding where AI should actually be used. Instead of allowing AI to make the core safety calculation, we separated the deterministic risk engine from the AI explanation layer. This made the system more transparent and allowed AI to focus on converting structured data into useful information for humans.
What We Learned
This project taught us that building on the cloud isn't just about knowing individual AWS services. The real challenge is understanding how those services work together to turn an idea into a working product.
BrajSahayak was our attempt to take a real problem from our region and turn it into a working cloud application using AWS.
Loading article