Solovea-101/FuelIQ

A mobile-first ML-powered fleet fuel efficiency optimization system designed for Kenya’s transport sector. Built with React Native (frontend), FastAPI (backend), SQLite (database), and machine learning (XGBoost, Random Forest). # FuelIQ A fleet fuel management system built for Kenya's transport industry. FuelIQ helps fleet managers track fuel consumption, monitor driver performance, and reduce operating costs through ML-powered predictions and real-time trip tracking. --- ## Features ### Driver App - Real-time GPS trip tracking with live waypoint recording - Fuel consumption predictions powered by a Gradient Boosting ML model - Personal performance dashboard (distance, fuel used, efficiency score) - Assigned vehicle details and service status - Offline trip queuing with automatic sync on reconnect ### Fleet Manager App - Fleet-wide analytics and fuel efficiency trends - Driver performance rankings and comparisons - Vehicle management (status, assignments, service scheduling) - Budget tracking by route - AI-generated fuel-saving recommendations ### Admin App - User and company management - System-wide analytics - Fuel price and settings configuration --- ## Tech Stack ### Mobile (this repo) | Layer | Technology | |-------|-----------| | Framework | React Native 0.81 + Expo SDK 52 | | Navigation | Expo Router 6 (file-based) | | Language | TypeScript (strict mode) | | Styling | NativeWind (Tailwind CSS for RN) | | State | Zustand | | Server state | TanStack Query v5 | | Animations | React Native Reanimated 4 | | Storage | Expo SecureStore | | Icons | Expo Vector Icons (Ionicons) | ### Backend (in `/backend`) | Layer | Technology | |-------|-----------| | Framework | FastAPI 0.115 | | Language | Python 3.11 | | Database | PostgreSQL (production) / SQLite (local dev) | | ORM | SQLAlchemy 2.0 + Alembic migrations | | Auth | JWT (python-jose) + bcrypt | | ML Model | Gradient Boosting Regressor (scikit-learn) | | Server | Uvicorn | | Deployment | Railway (Docker) | ### ML Model - **Algorithm:** Gradient Boosting Regressor - **Accuracy:** R² = 0.9922, MAE = 0.84 L, MAPE = 5.45% - **Features:** 13 inputs — engine size, cylinders, distance, speed, idle time, load weight, fuel type, route t …