Voyageur — AI Travel Safety Companion

A full-stack travel safety app that gives real-time crime risk analysis, hospital lookup, route planning, and an AI chat assistant — all in one dashboard.

---

What's Been Built

Backend (FastAPI + Python)

Auth System

• JWT-based auth with access tokens (30 min) and refresh tokens (7 days)
• POST /api/v1/signup — register and get tokens
• POST /api/v1/login — login and get tokens
• POST /api/v1/refresh — refresh access token
• POST /api/v1/logout — stateless logout
• GET /api/v1/me — get current user info
• Full CRUD for users (/api/v1/users)

Trip Management

• Authenticated CRUD for trips tied to the logged-in user
• POST /api/v1/trips — create trip
• GET /api/v1/trips — get all trips for current user
• GET /api/v1/trips/{id} — get specific trip
• PUT /api/v1/trips/{id} — update trip
• DELETE /api/v1/trips/{id} — delete trip

AI Chat

• POST /api/v1/ai/chat — conversational AI powered by Google Gemini
• Automatically detects Indian state from message context and injects crime safety data
• Falls back across multiple Gemini models (gemini-2.5-flash, 2.0-flash, 1.5-flash)
• Chat message persistence: POST/GET /api/v1/ai/chat-messages
• GET /api/v1/ai/crime-warning/{state} — AI-generated safety warning for a state

Travel & Crime Risk

• GET /api/v1/travel/crime-risk?lat=&lng= — crime risk for GPS coordinates
• GET /api/v1/travel/districts-in-state?lat=&lng= — all districts in detected state with risk data
• GET /api/v1/travel/districts — all districts risk data
• GET /api/v1/travel/hospitals?lat=&lng=&radius=&limit= — nearby hospitals via haversine distance

Data & Services

• Crime data loaded from processed_crime_data.csv with normalized risk scores (1–10 scale)
• Handles Telangana/Andhra Pradesh historical dataset split
• Reverse geocoding via Nominatim with coordinate-based fallback using district_centroids.json
• Hospital data from processed_hospital_directory.csv with coordinate parsing
• AI notebooks for crime and hospital data analysis (ai_notebooks/)

Database

• PostgreSQL via Supabase, SQLAlchemy ORM
• Models: User, Trip, ChatMessage

---

Frontend (React + Vite + Tailwind)

Pages

• / — Home/landing page with login/signup navigation
• /login — Animated login form with framer-motion
• /signup — Animated signup form
• /dashboard — Main dashboard (protected)
• /chat — Full-page AI chat (protected)

Dashboard Features

• Dark glassmorphism UI with purple/blue gradient theme
• Live interactive map (Leaflet + OpenStreetMap) showing your current location
• Crime risk circle overlay around your position (color-coded: red/yellow/green)
• Click anywhere on the map to get crime risk for that location
• Route planner with From/To inputs, map pin picker, and "Current Location" shortcut
• Turn-by-turn directions panel powered by OSRM routing API
• Estimated travel time and distance summary
• Nearby hospitals toggle with green markers and 30km radius overlay
• Risk panel sidebar showing live location risk + clicked location risk
• Floating AI chat bubble (bottom-right) with full conversation history

AI Chat Page

• Full-screen chat interface with message history
• Sends user's GPS coordinates as trip context to the AI
• Typing indicator animation
• Back-to-dashboard navigation

Auth & State

• AuthContext with login, signup, logout, and token persistence in localStorage
• axiosInstance with automatic token refresh interceptor (queues failed requests during refresh)
• ProtectedRoute component redirects unauthenticated users to /login
• useLocation hook for real-time GPS via browser Geolocation API

Deployment

• Frontend deployed on Vercel (voyageur-sable.vercel.app)
• Backend deployed on Render (voyageur-1i0h.onrender.com)

---

Tech Stack

Layer	Tech
Frontend	React 18, Vite, Tailwind CSS, Leaflet, framer-motion, Axios
Backend	FastAPI, SQLAlchemy, Pydantic, bcrypt, python-jose
AI	Google Gemini API (google-genai)
Database	PostgreSQL (Supabase)
Routing	OSRM (open-source routing)
Geocoding	Nominatim (OpenStreetMap)
Data	Pandas, crime + hospital CSVs

---

Running Locally

Backend

cd Backend
python -m venv venv
venv\Scripts\activate        # Windows
pip install -r requirements.txt

Create Backend/.env:

DATABASE_URL=your_supabase_postgres_url
SECRET_KEY=your_jwt_secret
GEMINI_API_KEY=your_google_gemini_key

uvicorn app.main:app --reload

API runs at http://localhost:8000. Docs at http://localhost:8000/docs.

Frontend

cd Frontend/client
npm install

Create Frontend/client/.env.local:

VITE_API_URL=http://localhost:8000

npm run dev

App runs at http://localhost:5173.