title	FitAI
emoji	💪
colorFrom	blue
colorTo	green
sdk	docker
app_port	8080

FitAI - AI Fitness Coach PWA

FitAI is a full-stack Progressive Web App that provides personalized fitness coaching powered by Google Gemma AI. Users can log workouts and health metrics, then chat with an AI coach that gives context-aware advice based on their actual recorded data.

Features

• AI Chat Coach - Conversational fitness coaching powered by Gemma 3 (1B-IT) via Google GenAI, with RAG-based context retrieval from user data
• Coach Personas - Choose between strict, encouraging, or analytical coaching styles
• Workout Logging - Track workouts with exercises, sets, reps, weight, and distance
• Health Metrics - Log weight, sleep hours, and energy levels
• JWT Authentication - Secure registration and login with hashed passwords
• Chat History - Conversation continuity with recent message recall

Roadmap & Architecture Evolution

This project is currently undergoing a major architectural refactoring to ensure data precision and system scalability.

• Phase 1: Deterministic Context (Current) - Implementing strict SQLite querying for structured health metrics to guarantee 100% accuracy in the LLM context window.
• Phase 2: Unstructured Data Embedding (WIP) - Integrating Qdrant Vector Database for semantic search over unstructured workout notes.
• Phase 3: Agentic Routing - Deploying an Intent Router to dynamically route queries between standard SQL database (for exact stats) and Qdrant (for semantic advice).

Tech Stack

Layer	Technology
Frontend	React 19, TypeScript, Tailwind CSS 4, Vite 8, React Router 7
Backend	FastAPI, SQLAlchemy, Pydantic, PyJWT, bcrypt
AI	Google GenAI SDK (Gemma 3 1B-IT)
Database	SQLite
Deployment	Docker, Docker Compose, Nginx

Project Structure

.
├── backend/
│   ├── app/
│   │   ├── main.py            # FastAPI app entry point
│   │   ├── auth.py            # JWT & password hashing utilities
│   │   ├── database.py        # SQLAlchemy engine & session
│   │   ├── models.py          # ORM models (User, Workout, Exercise, HealthMetric, ChatHistory)
│   │   ├── schemas.py         # Pydantic request/response schemas
│   │   ├── routers/
│   │   │   ├── auth.py        # /auth - register, login, profile, persona
│   │   │   ├── workouts.py    # /workouts - CRUD
│   │   │   ├── health.py      # /health - CRUD
│   │   │   └── chat.py        # /chat - AI conversation
│   │   └── services/
│   │       └── rag_service.py # Builds user context for AI prompts
│   ├── tests/
│   ├── Dockerfile
│   └── requirements.txt
├── frontend/
│   ├── src/
│   │   ├── pages/             # Login, Register, Dashboard, WorkoutLog, HealthLog, ChatPage
│   │   ├── components/        # Layout
│   │   └── api/               # API client
│   ├── Dockerfile
│   └── package.json
├── docker-compose.yml         # Production multi-container setup
├── Dockerfile.dev             # Development environment image
└── run_dev.sh                 # Dev container launch script

Getting Started

Prerequisites

• Docker & Docker Compose
• A Google AI API key for the Gemma model

Environment Variables

Create backend/.env:

GOOGLE_API_KEY=your_google_api_key_here
JWT_SECRET=your_jwt_secret_here

Run with Docker Compose (Production)

docker compose up --build

• Frontend: http://localhost:8080
• Backend API: http://localhost:8000
• Health check: http://localhost:8000/health

Local Development

Backend:

cd backend
python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
uvicorn app.main:app --reload

Frontend:

cd frontend
npm install
npm run dev

Development Container

A full dev container with Python 3.14, Node.js 24, and AI CLI tools is provided:

./run_dev.sh
sudo docker exec -it fitai-dev-env /bin/bash

API Endpoints

Method	Endpoint	Description
POST	/auth/register	Register a new user
POST	/auth/login	Login and receive JWT
GET	/auth/me	Get current user profile
PATCH	/auth/me/persona	Update coach persona
POST	/workouts	Create a workout
GET	/workouts	List user's workouts
POST	/health	Create a health metric
GET	/health	List user's health metrics
POST	/chat	Send a message to the AI coach
GET	/health	Health check

License

This project is private and not licensed for redistribution.