AI-powered workout generation • Precision nutrition • Real-time biometric telemetry
ApexAI is a full-stack, AI-integrated fitness platform that delivers personalized workout protocols, precision nutrition targets, and an intelligent chat assistant — all wrapped in a premium, military-grade dark UI.
It uses OpenAI GPT-4o-mini (with a local Llama 3 fallback via Ollama) to generate adaptive plans based on your biometric profile: age, gender, weight, height, and fitness goals.
Secure authentication with JWT tokens and a multi-theme Visual Spectrum selector
AI-calculated calorie and macro targets with dietary guidance
Context-aware fitness assistant with image-based meal analysis (Neural Vision Protocol)
| Feature | Description |
|---|---|
| AI Workout Plans | Personalized 4-week progressive overload protocols generated by GPT-4o-mini |
| Precision Nutrition | Calorie and macronutrient targets calculated from your biometric profile |
| AI Chat Assistant | Context-aware fitness Q&A with multi-turn conversation memory |
| Neural Vision | Snap a photo of your meal → instant nutritional analysis |
| Progress Tracking | Log workouts, track weight, and visualize biometric trends over time |
| Multi-Theme UI | 3 premium visual spectrums: Apex Core, Hazard State, Void Protocol |
| Biometric Sync | Integration bridge for Garmin, Whoop, and Apple Health devices |
| Stealth HUD | Non-intrusive tactical status overlay with auto-fade |
| Layer | Technology |
|---|---|
| Frontend | Next.js 14, React, TypeScript, Tailwind CSS |
| Backend | FastAPI (Python), SQLAlchemy ORM |
| Database | SQLite (dev) / PostgreSQL (prod) |
| AI Engine | OpenAI GPT-4o-mini + Llama 3 (Ollama fallback) |
| Auth | JWT (JSON Web Tokens) with bcrypt hashing |
| State | Zustand (high-frequency biometric telemetry) |
| Deployment | Docker + Docker Compose |
- Node.js 18+
- Python 3.10+
- OpenAI API Key (get one here)
git clone https://github.com/YOUR_USERNAME/ApexAI.git
cd ApexAICreate a .env file in the project root:
DATABASE_URL=sqlite:///./apex.db
OPENAI_API_KEY=your_openai_api_key_here
SECRET_KEY=your_jwt_secret_key
ALLOWED_ORIGINS=http://localhost:3000
OLLAMA_URL=http://localhost:11434/api/chatcd backend
pip install -r requirements.txt
cd ..
python3 -m uvicorn backend.main:app --reload --port 8000cd frontend
npm install
npm run devNavigate to http://localhost:3000 and register a new account.
docker-compose up --buildThis will spin up both the frontend and backend services automatically.
ApexAI/
├── frontend/ # Next.js 14 application
│ ├── src/
│ │ ├── app/ # App router pages (dashboard, login, etc.)
│ │ ├── components/ # Reusable UI components
│ │ ├── context/ # React context providers
│ │ ├── store/ # Zustand state management
│ │ └── lib/ # API client and utilities
│ └── public/ # Static assets (favicon, logo)
│
├── backend/ # FastAPI application
│ ├── routers/ # API route handlers
│ ├── models.py # SQLAlchemy database models
│ ├── schemas.py # Pydantic validation schemas
│ ├── auth.py # JWT authentication logic
│ └── database.py # Database engine configuration
│
├── ai/ # AI recommendation engine
│ └── recommender.py # GPT-4o-mini + Llama 3 integration
│
├── docs/screenshots/ # Application screenshots
├── docker-compose.yml # Container orchestration
└── .env # Environment configuration
User Request
│
▼
┌─────────────┐ ┌──────────────┐ ┌─────────────────┐
│ FastAPI │────▶│ GPT-4o-mini │────▶│ Structured │
│ Router │ │ (Primary) │ │ JSON Response │
└─────────────┘ └──────────────┘ └─────────────────┘
│ fail
▼
┌──────────────┐
│ Llama 3 │
│ (Ollama) │
└──────────────┘
│ fail
▼
┌──────────────┐
│ Static │
│ Fallback │
└──────────────┘
The system uses a 3-tier fallback strategy: GPT-4o-mini → Llama 3 (local) → Static Elite Protocol. This ensures the platform never shows an empty state, even when all AI services are offline.
This project is licensed under the MIT License — see the LICENSE file for details.
Engineered by Arjun · v5.4 // Neural Collective
Built with passion and way too much caffeine