Full-stack AI wardrobe platform spanning the web, a REST API, and native Android.
Live app · Engineering docs · Report an issue
FitGPT helps users organize a digital wardrobe and receive explainable outfit recommendations grounded in their preferences, clothing inventory, weather, time, and wear history. The product combines a React experience, a FastAPI service, persistent data, and a Kotlin Android client while preserving local-first fallbacks for core flows.
| Area | Implementation |
|---|---|
| Web | React, React Router, Three.js, TensorFlow.js, Recharts |
| API | FastAPI, SQLAlchemy, JWT authentication, Google OAuth |
| Data | SQLite or PostgreSQL with users, wardrobe items, outfits, plans, and feedback |
| AI and context | Groq-backed recommendations, AURA assistant, MobileNet tagging, weather context |
| Mobile | Kotlin, Jetpack Compose, Retrofit |
| Quality | 185+ backend tests and 617 web tests; CI runs tests and a production web build |
| Delivery | Vercel, Render, and GitHub Actions |
- Digital wardrobe CRUD, bulk upload, filtering, duplicate detection, and item reuse insights
- Explainable outfit recommendations scored for style, fit, color, season, weather, and history
- AURA conversational styling assistant with deterministic fallback behavior
- Weekly outfit planning and forecast-aware trip packing
- Saved outfits, favorites, wear history, and preference-based feedback
- 3D outfit preview and a drag-and-drop builder
- Secure account flows with JWT and Google OAuth
- Responsive web UI plus a native Android client
flowchart LR
Web[React web] --> API[FastAPI service]
Android[Android app] --> API
API --> DB[(SQLite or PostgreSQL)]
API --> AI[Groq and local scoring]
API --> Weather[Weather data]
Core wardrobe workflows degrade gracefully to browser storage when the backend is unavailable. Server-backed authentication, synchronization, and AI enhancements layer on top of that local-first experience.
FitGPT/
├── backend/ FastAPI application, data models, AI services, and tests
├── web/ React wardrobe experience and test suite
├── app/ Native Kotlin / Jetpack Compose client
├── docs/ Product, architecture, API, and release documentation
└── .github/workflows Continuous integration
cd backend
python -m venv .venv
source .venv/bin/activate # Windows: .venv\Scripts\activate
pip install -r requirements.txt
cp .env.example .env
uvicorn app.main:app --reloadConfigure the values documented in backend/.env.example, including the database and any optional external-service credentials.
cd web
npm install
npm run devOpen the repository root in Android Studio, use an Android SDK compatible with API 36, and run the app configuration on an emulator or device (minimum API 26).
cd backend && pytest
cd web && npm run test:ci
cd web && npm run buildGitHub Actions runs the backend suite on Python 3.12 and the frontend checks on Node 20.
AI-generated outfit suggestions are recommendations, not ground truth. Production use should preserve the current explainability and fallback paths while adding model-output monitoring, clear data-retention controls, user correction mechanisms, and careful handling of uploaded wardrobe images and preference data.
Built by Muhammad Imran — portfolio · LinkedIn