AI & Full Stack · 2024
AASRA AI System
An intelligent disaster management ecosystem designed to save lives through real time early warnings.
+35%
Warning Accuracy
1,000+
Simulated Users
Overview
AASRA is a complete disaster management ecosystem built from the ground up to handle critical emergency response. It consists of three main pillars: an intuitive Android application for citizens, a React powered command center for administrators, and an intelligent Python backend. Together, they facilitate immediate crisis reporting, verify the authenticity of disaster imagery using machine learning, and allow authorities to coordinate volunteer efforts and resource allocation instantly.
The Problem
Disaster response in localized regions often suffers from delayed reporting, fraudulent alerts, and inefficient resource allocation during critical, time sensitive emergencies.
The Solution
I engineered a tri part ecosystem utilizing Supabase and Firebase for real time data synchronization. A native Android app allows citizens to report emergencies using OSMDroid maps. A React dashboard gives authorities real time control. Crucially, a PyTorch powered AI engine verifies images using MobileNetV2 and detects fraud using an Isolation Forest model, ensuring only genuine emergencies are escalated.
Key Features
What makes it tick.
AI Image Verification
Integrated MobileNetV2 model to automatically verify the authenticity of user uploaded disaster images.
Spam & Fraud Detection
Utilizes an Isolation Forest model to identify abnormal reporting frequencies and flag malicious accounts.
Real Time Sync & Offline Capability
Ensures seamless communication and offline first capabilities for over 1000 simulated concurrent users.
Tech Stack
Built with precision.
Challenges
Integrating complex PyTorch machine learning models into a highly concurrent FastAPI backend while ensuring the real time synchronization with Firebase remained virtually instantaneous during high load simulated disasters.
Learnings
Gained massive insight into architectural design for distributed systems. Learned how to successfully bridge native mobile applications, web portals, and complex AI microservices into one cohesive, highly available ecosystem.
More Work
Related Projects.
Interested in working together?
Let's build something remarkable.