AI Platform

Full-stack AI chat platform with intelligent model routing built with Flask, Tailwind CSS, and open-source AI models.

Features

🤖 Intelligent Model Routing

• DeepSeek Coder: Specialized for coding, debugging, and programming tasks
• Llama.cpp: Optimized for document processing, PDFs, and large files
• Vicuna: Multimodal AI for images, videos, and rich media content
• GPT4All: General-purpose conversational AI for everyday tasks
• Auto-Select: Automatically routes queries to the best model based on content type

✨ Core Features

• User authentication with registration, login, and password reset
• Rate limiting by tier (Free: 10/hour, Premium: 100/hour, Admin: 1000/hour)
• Chat history with message persistence
• Admin panel for user management and analytics
• Payment integration with Midtrans
• Email notifications via Postfix SMTP
• Dark/light mode toggle
• Markdown rendering with code syntax highlighting
• RESTful API for programmatic access

🛡️ Security

• CSRF protection
• Secure password hashing (Werkzeug)
• Session management with secure cookies
• Rate limiting (Redis-backed)
• Input sanitization
• Security headers (CSP, X-Frame-Options, etc.)

Quick Start

Using Docker (Recommended)

# Clone repository
git clone https://github.com/Bucin404/AI-platform.git
cd AI-platform

# Configure environment
cp .env.example .env
# Edit .env with your configuration

# Download & integrate AI models (AUTOMATIC - recommended!)
python download_models.py
# Downloads models + installs dependencies + configures platform
# Use --lite for smaller models (~1.6GB vs ~16GB)

# Start services
docker-compose up -d

# Initialize database
docker-compose exec web flask db upgrade

# Create admin user
docker-compose exec web flask create-admin

# Access at http://localhost:5000

Using Docker on MacBook M4 (Apple Silicon)

# Use M4-optimized configuration
docker-compose -f docker-compose.m4.yml up -d

# Initialize database
docker-compose -f docker-compose.m4.yml exec web flask db upgrade

# Create admin user
docker-compose -f docker-compose.m4.yml exec web flask create-admin

# See full M4 setup guide: docs/SETUP_M4.md

Local Development (Without Docker)

Important: For local development, use .env.local.example instead of .env.example:

# Clone and setup
git clone https://github.com/Bucin404/AI-platform.git
cd AI-platform

# Create virtual environment
python3 -m venv venv
source venv/bin/activate

# Install dependencies
pip install -r requirements.txt

# Configure for LOCAL development
cp .env.local.example .env
# Edit .env - use localhost for database, or SQLite for simplicity

# Option 1: Use SQLite (no PostgreSQL needed)
# In .env: DATABASE_URL=sqlite:///app.db

# Option 2: Use local PostgreSQL
# Install PostgreSQL, create database, then:
# In .env: DATABASE_URL=postgresql://user:pass@localhost:5432/aiplatform

# Initialize database
export FLASK_APP=run.py
flask db init
flask db migrate -m "Initial migration"
flask db upgrade

# Download AI models (automatic integration!)
python download_models.py

# Run application
python run.py

# Access at http://localhost:5000

Troubleshooting? See docs/TROUBLESHOOTING.md for common issues and solutions.

Previous Local Development

# Create virtual environment
python3 -m venv venv
source venv/bin/activate

# Install dependencies
pip install -r requirements.txt

# Set up environment
cp .env.example .env
# Edit .env with your configuration

# Download & integrate AI models (AUTOMATIC!)
python download_models.py --lite  # Lite version for local dev (~1.6GB)
# Or use: python download_models.py  # Full version (~16GB)
# This automatically installs deps and configures the platform!

# Initialize database
export FLASK_APP=run.py
flask db init
flask db migrate -m "Initial migration"
flask db upgrade

# Create admin user
flask create-admin

# Run application
python run.py

Documentation

• Installation Guide - Detailed installation instructions
• MacBook M4 Setup - Optimized setup for Apple Silicon (NEW)
• Model Download Guide - How to download and configure AI models
• Deployment Guide - Production deployment on Ubuntu
• API Documentation - REST API reference

Tech Stack

Backend:

• Flask (Python web framework)
• PostgreSQL (Database)
• Redis (Caching & rate limiting)
• SQLAlchemy (ORM)
• Flask-Login (Authentication)
• Flask-WTF (Forms & CSRF)
• Flask-Mail (Email)
• Flask-Limiter (Rate limiting)
• Flask-Migrate (Database migrations)

Frontend:

• Tailwind CSS (Styling)
• Vanilla JavaScript
• Marked.js (Markdown rendering)
• Highlight.js (Code syntax highlighting)

AI Models:

• DeepSeek Coder
• Llama.cpp
• Vicuna
• GPT4All

Infrastructure:

• Docker & Docker Compose
• Nginx (Reverse proxy)
• Postfix (Email)
• Midtrans (Payments)

Project Structure

AI-platform/
├── app/
│   ├── blueprints/          # Flask blueprints
│   │   ├── auth/            # Authentication routes
│   │   ├── chat/            # Chat routes
│   │   ├── admin/           # Admin panel routes
│   │   ├── payments/        # Payment routes
│   │   └── api/             # API routes
│   ├── models/              # Database models
│   ├── services/            # Business logic services
│   │   ├── model_service.py # AI model adapters
│   │   ├── email_service.py # Email service
│   │   └── payment_service.py # Payment service
│   ├── templates/           # HTML templates
│   ├── static/              # Static files (CSS, JS)
│   └── utils/               # Utility functions
├── tests/                   # Unit tests
├── docs/                    # Documentation
├── config.py                # Configuration
├── run.py                   # Application entry point
├── requirements.txt         # Python dependencies
├── Dockerfile               # Docker configuration
└── docker-compose.yml       # Docker Compose configuration

Environment Variables

See .env.example for all required environment variables. Key variables:

# Flask
SECRET_KEY=your-secret-key
DEBUG=False

# Database
DATABASE_URL=postgresql://user:pass@localhost:5432/aiplatform

# Redis
REDIS_URL=redis://localhost:6379/0

# Email (Postfix)
MAIL_SERVER=localhost
MAIL_PORT=587
MAIL_USERNAME=
MAIL_PASSWORD=

# Midtrans Payment
MIDTRANS_SERVER_KEY=your-server-key
MIDTRANS_CLIENT_KEY=your-client-key
MIDTRANS_IS_PRODUCTION=False

# Admin
ADMIN_EMAIL=admin@aiplatform.com
ADMIN_PASSWORD=changeme

Testing

# Run all tests
pytest

# Run with coverage
pytest --cov=app tests/

# Run specific test file
pytest tests/test_auth.py

API Usage

import requests

# Login
session = requests.Session()
session.post('http://localhost:5000/auth/login', data={
    'email': 'user@example.com',
    'password': 'password'
})

# Send chat message with auto model selection
response = session.post('http://localhost:5000/api/chat', json={
    'message': 'Write a Python function to sort a list',
    'model': 'auto'  # Will automatically use DeepSeek for coding
})

print(response.json())

Model Routing Logic

The platform intelligently routes queries based on content detection:

Query Type	Model	Example Keywords
Coding	DeepSeek Coder	code, function, debug, python, javascript
Documents	Llama.cpp	pdf, document, file, csv
Images/Videos	Vicuna	image, photo, video, analyze
General	GPT4All	weather, explain, tell me

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

MIT License - see LICENSE file for details

Support

• GitHub Issues: https://github.com/Bucin404/AI-platform/issues
• Email: support@aiplatform.com

Credits

Built with ❤️ using open-source AI models and modern web technologies.