tursi 0.1.0

A simple framework to deploy AI models locally with one command, no containers needed

tursi-ai

A simple, lightweight framework to deploy AI models locally with a single command—no Docker, no external services required.

Overview

tursi-ai lets you run AI models (like text classification) on your machine with minimal setup. Our unique selling proposition: "AI deployment, one command, no containers needed." The base install is ~150-200MB, with an additional ~250MB for the default model on first run. Perfect for developers who want simplicity without complexity.

Features

• One-command deployment: Start a model server with a single script.
• No containers: Runs directly in your Python environment.
• Lightweight: Minimal dependencies, small footprint.
• Extensible: Built for easy customization and growth.

Getting Started

Prerequisites

• Python 3.8+ (tested with 3.12)
• A virtual environment (recommended)

Installation

1. Clone or download this project (for now, until PyPI packaging):

   git clone <path-to-your-local-repo>  # Or copy the folder manually
   cd tursi-ai
   

2. Set up a virtual environment:

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate
   

3. Install dependencies:

   pip install -r requirements.txt
   

   This installs everything needed for both tursi-engine and tursi-test.

Usage

1. Deploy a model: Run the engine to start a Flask server with the default model (distilbert-base-uncased-finetuned-sst-2-english):

   python tursi-engine/tursi-engine.py
   

   Output:

   Loading model...
   Model loaded!
   Deploying at http://localhost:5000/predict
   

2. Test the deployed model: Use the included tursi-test script to send a prompt to the server. In a separate terminal (with the virtual environment activated):

   python tursi-test/tursi-test.py --prompt "I love AI"
   

   Expected output:

   {
     "label": "POSITIVE",
     "score": 0.999...
   }
   

   Alternatively, test with curl:

   curl -X POST -H "Content-Type: application/json" -d '{"text":"I love AI"}' http://localhost:5000/predict
   

Project Structure

tursi-ai/
├── .github/          # GitHub Actions (e.g., linting)
├── examples/         # Sample configs (future use)
├── tursi-engine/     # Core deployment script
├── tursi-test/       # Testing utility
├── LICENSE           # MIT License
├── README.md         # This file
├── requirements.txt  # Dependencies
└── .gitignore        # Git ignore rules

Roadmap

• Add CLI support (e.g., tursi-engine up --model <model-name>).
• Package as a PyPI module (pip install tursi).
• Support more model types beyond text classification.

Contributing

This is an open-source project under the MIT License. Feel free to fork, tweak, or submit ideas! For now, the repo is local—stay tuned for a public release.

License

MIT License—see LICENSE for details.

Acknowledgments

Built with ❤️ using:

• Transformers by Hugging Face
• Flask
• PyTorch

Built by BlueTursi.