Data Genie: Synthetic Data Generator

This project generates synthetic data for testing, prototyping, and analytics. It provides a web interface (frontend in React, backend in Flask) and supports custom data schemas, LLM-powered data generation (Ollama, OpenAI, Azure OpenAI), and export to CSV/JSON.

Features

• Web frontend built with React (see frontend/)
• Flask backend for API and LLM integration
• Customizable data schemas
• LLM-powered synthetic data generation (Ollama, OpenAI, Azure OpenAI)
• Example report output (report.html)

Quick Start

1. Clone this repo and enter the folder:

   git clone https://github.com/simagix/data-genie.git
   cd data-genie

2. Create and activate a virtual environment:

   python3 -m venv venv
   source venv/bin/activate

3. Install dependencies:

   pip install -r requirements.txt

4. (Optional) Install Ollama and pull a model for LLM-powered generation (see below).

5. Run the backend:

   python app.py

6. Run the frontend:

   cd frontend
   npm install
   npm start

   If you see dependency errors (ERESOLVE), run:

   npm install --legacy-peer-deps

   This will skip strict peer dependency checks and resolve most install issues with older or conflicting packages.

LLM Backends

• Ollama (local):
  • Install via Homebrew (macOS):

    brew install ollama
    ollama pull mistral:7b-instruct

  • Set environment variables (optional):

    LLM_BACKEND=ollama
    OLLAMA_URL=http://localhost:11434/api/generate
    OLLAMA_MODEL=mistral:7b-instruct

• OpenAI (cloud):
  • Set environment variable:

    OPENAI_API_KEY=sk-...
    OPENAI_MODEL=gpt-4o

• Azure OpenAI (cloud):
  • Set environment variables:

    LLM_BACKEND=azure
    AZURE_OPENAI_ENDPOINT=https://<resource>.openai.azure.com/
    AZURE_OPENAI_API_KEY=...
    AZURE_OPENAI_DEPLOYMENT=gpt-4
    AZURE_OPENAI_API_VERSION=2024-12-01-preview

Run the App

Start Flask backend:

python app.py

Start React frontend:

cd frontend
npm start

By default, frontend runs at http://localhost:3000 and backend at http://localhost:5000

Usage

1. Open the frontend in your browser.
2. Define a data schema or use a preset.
3. Generate synthetic data (optionally using LLM).

Notes

• LLM features require Ollama, OpenAI, or Azure OpenAI and a supported model.