🌱 SynapseAid — Emergency Response & Multi-Agent Dispatch Simulation Prototype

SynapseAid is an interactive, browser-based emergency-response simulation prototype implementing deterministic multi-agent coordination (Gale-Shapley matching), client-side hybrid retrieval (dense vector + BM25 ranking), and optional LLM integration for disaster scenarios.

🌐 Live Demo: https://synapse-aid.vercel.app

---

📌 Project Scope & Technical Implementation Transparency

To ensure complete interview defensibility and clarity regarding what is built from scratch vs. external/simulated:

System Layer	Technical Implementation Details	Implementation Scope
Gale-Shapley Matcher	Stable-matching algorithmic solver prioritizing role affinities, distance matrices, certifications, and responder score ratings	Implemented from scratch (TypeScript)
Dense + Sparse RAG Engine	Client-side 24-D synthetic semantic vector projection + BM25 keyword tokenizer with rank fusion and XAI explainability chips	Implemented from scratch (TypeScript)
Deterministic Audit Ledger	Cryptographic SHA-256 block receipt generator creating immutable verification trails for disaster manifests	Implemented from scratch (TypeScript)
GIS Tactical Map & Vectors	Interactive relief map, dynamic incident bounding boxes, and transit vector rendering	Implemented from scratch (Canvas / DOM)
Speech & NLU Processing	Real-time speech-to-text distress capture and speech synthesis via Web Speech API (with optional Google Gemini API fallback)	Browser API & Optional External LLM
Emergency Telemetry & Depletion	Synthetic multi-agent response timings (~97ms latency models) and regional warehouse depletion rate simulations	Deterministic Simulation Layer

Note

Zero Mandatory Cloud Dependencies: The system executes 100% deterministically in the client browser out-of-the-box, with zero external database or paid API requirements.

---

⚡ System Architecture & Pipeline

📡 Civilian Distress Signal (Voice SOS / NLP Report)
       │
       ▼
 🤖 NLP Entity Triage & Severity Classification (Critical / High / Medium / Low)
       │
       ▼
 📚 Hybrid Vector RAG Retrieval (24-D Dense Cosine + BM25 Sparse Overlap)
       │
       ▼
 🧠 Multi-Agent Coordination Mesh (Simulated)
       ├── 📦 Logistics & Route Router (Depot buffer allocation & transit vectors)
       ├── 👥 Gale-Shapley Matcher (Role affinity, proximity, mission rating)
       └── 🛡️ Audit Ledger (SHA-256 manifest block receipt hashing)
       │
       ▼
 🚨 Operations GIS Tactical Map & Mission Manifest PDF Export

---

🚀 Key Modules

1. 🎙️ Field Coordination Desk & Voice SOS

• Web Speech API: Real-time Speech-to-Text (STT) distress capture and Text-to-Speech (TTS) audio dispatch playback (English, Hindi, Spanish).
• Simulation Scenarios: One-click test scenarios for Riverside Flood Inundation, Building Collapse, and Shelter Evacuation.
• Collapsible Layout: Left coordination desk can be collapsed ([◀ Hide Assistant]) to expand the GIS studio to full screen.

2. 🗺️ Tactical Operations GIS Relief Map

• Bounded Layout: Clean separation between the interactive map canvas and the dedicated incident operations panel below.
• Map Elements:
  • District / sector boundaries (Sector Alpha Industrial, Sector Bravo Riverbank, Sector Gamma Highland, Sector Delta Residential).
  • Translucent flood inundation zone polygon styling.
  • Safe Shelters (🟢 with live capacity tracking).
  • Tactical Supply Depots (🔵 with VHF radio channels).
  • Field Responders (🟣 with deployment tracking).
  • Pulsing deployment route vectors and interactive map legend.

3. 🧠 Hybrid Vector RAG Protocol Studio

• Dense + Sparse Rank Fusion: Combines 24-D semantic projection with BM25 lexical overlap.
• Explainable AI (XAI): "Why this match?" keyword tags (✓ Flash flood, ✓ Water purification) highlight retrieval rationale.
• Technical Breakdown: Displays real-time Semantic Cosine %, Keyword BM25 %, and Final Hybrid Score %.
• Protocol Inspector: In-memory document chunker and full-text viewer for NDRF, FEMA, and WHO crisis manuals.

4. 👥 Gale-Shapley Volunteer Matching Engine

• Stable Matchmaker Algorithm: Matches field responders based on role priority, distance matrices, certifications, and past mission ratings.
• Dynamic AI Match Scores: Real-time progress bars (⭐ 98% Match, ⭐ 94% Match) calculated against active incident requirements.

5. 📊 Telemetry & Resource Depletion Analytics Hub

• Hourly Supply Burn Curve: Area chart with 1H | 6H | 24H time-range selector.
• Sub-Agent Latency Waterfall: Visualizes simulated pipeline execution latency across agents.
• KPI Metrics: Resource burn velocity (431 units/hr), priority incident ratio, and warehouse buffer allocations.

6. 📄 Cryptographic Manifest & Mission PDF Export

• Generates SHA-256 block receipts and exports printable, high-resolution mission PDF reports using jspdf and html2canvas.

---

🛠️ Tech Stack

• Frontend: React 18, TypeScript, Vite
• Styling: Tailwind CSS v4, Custom CSS Design System
• Data Visualization: Recharts, Canvas Confetti
• Icons: Lucide React
• Document Generation: jsPDF, html2canvas, DOMPurify
• AI & NLP: Client-side synthetic vector projection, BM25 tokenizer, optional Google Gemini API

---

📦 Getting Started

Prerequisites

• Node.js: v18.0.0 or higher
• npm: v9.0.0 or higher

Installation

# Clone the repository
git clone https://github.com/lakshiaharan/synapse-aid.git
cd synapse-aid

# Install dependencies
npm install

# (Optional) Configure Gemini API Key
cp .env.example .env
# Add your Gemini key in .env if desired: VITE_GEMINI_API_KEY=your_key_here

# Start local development server
npm run dev

Visit http://localhost:5173/ in your browser.

Production Build

# Build production bundle
npm run build

# Preview production build locally
npm run preview

---

☁️ Deployment (Vercel)

This project is configured for one-click deployment on Vercel:

1. Fork or push this repository to your GitHub account.
2. Import the project in Vercel.
3. Build Settings:
   • Framework Preset: Vite
   • Build Command: npm run build
   • Output Directory: dist
4. (Optional) Add VITE_GEMINI_API_KEY under Environment Variables.
5. Click Deploy.

---

📄 License

MIT License © 2026 Lakshi Haran. See LICENSE for details.