Skip to content

CalmaCodes/Trahi

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

TRAHI – Multi-Agent AI Emergency Response System

🚨 Overview

TRAHI is an agentic AI-powered emergency coordination system designed to reduce life-threatening delays in India’s emergency response infrastructure.

It autonomously receives emergency calls in multiple Indian languages, understands urgency through voice analysis, selects the optimal responders based on real-time availability and hospital capacity, dynamically reroutes through traffic, and coordinates multi-agency response — with intelligent human fallback when required.

TRAHI transforms fragmented emergency workflows into a real-time, closed-loop AI system.


❗ Problem Statement

India loses over 150,000 lives annually due to delayed and fragmented emergency response.

Current systems rely heavily on:

  • Manual human dispatch
  • Static routing
  • Disconnected hospital databases
  • Language barriers
  • Poor inter-agency coordination

These delays cost critical minutes during the Golden Hour.


💡 Solution

TRAHI introduces an autonomous agentic layer between citizens and emergency services.

It performs:

Voice Intake → Severity Detection → Resource Matching → Hospital Bed Booking → Traffic-Aware Routing → Multi-Agency Coordination → Continuous Learning

All within seconds.

Humans are involved only when AI confidence is low.


🧠 Core Features

  • Multi-language AI voice intake with panic detection
  • Intelligent responder selection (availability + equipment + traffic + hospital beds)
  • Automatic hospital bed reservation
  • Dynamic rerouting with traffic prediction
  • Multi-agency coordination (Police + Fire + Medical)
  • Intelligent human-in-the-loop fallback
  • Continuous post-incident learning

🏗 Architecture

TRAHI follows a layered, multi-agent architecture:

  • Intake Layer (Voice + Text + IoT)
  • Orchestration Layer (Central Coordinator)
  • Agent Layer (Matcher, Router, Hospital, Multi-Agency, Learning)
  • Data Layer (Incident, Resource, Audit Logs)
  • Integration Layer (Maps, Hospital Systems, Emergency Services)
  • Authority Dashboard

Closed-loop operation: Sense → Decide → Act → Monitor → Re-optimize


🛠 Technology Stack (Prototype)

Backend: Python (FastAPI)
Frontend: React
Database: PostgreSQL + Redis
AI/NLP: Transformer-based NLP + Speech-to-Text
Routing: Google Maps / Mapbox API
Notifications: SMS Gateway
Deployment: Docker + Cloud Hosting


📊 Success Metrics

  • Reduced call-to-dispatch time
  • Ambulance ETA reduction
  • Reduced hospital turn-away rate
  • Golden Hour survival improvement
  • Lower dispatcher workload

🔐 Notice

This repository is shared for hackathon evaluation purposes only.

All rights reserved by the author.
No part of this project (including ideas, architecture, or documentation) may be copied, reused, modified, or commercialized without explicit written permission.


📌 Status

Hackathon Design Submission – Prototype in development.

About

TRAHI is an agentic AI-powered unified emergency coordination system that acts as an intelligent autonomous layer between citizens and emergency services (police, fire, medical).

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors