mlinit-cli 0.1.1

Production-ready ML project scaffolder

mlinit 🚀

The production-ready scaffolder for Machine Learning projects.

mlinit is a CLI tool designed to bootstrap Machine Learning projects with industry-standard best practices. It goes beyond simple directory creation by setting up a complete development environment with Poetry, Hydra, Git, and CI/CD out of the box.

---

💡 Why mlinit?

There are many scaffolders out there. Here is why mlinit is different:

Feature	Cookiecutter	Kedro	mlinit
Philosophy	Generic templating engine	Full-fledged framework	Lightweight scaffolder + Modern Stack
Config	Static JSON	Custom ConfigLoader	Hydra (Composable & Dynamic)
Deps	requirements.txt usually	pip-tools	Poetry (Native Integration)
Learning Curve	Low	High	Low (It's just a folder structure)

mlinit gives you the structure of a framework without forcing you to learn a new API. It sets up the tools you already love (Hydra, Poetry, PyTorch) so you can just write code.

✨ Features

• 🏗 Production-Grade Structure: Generates a robust directory layout optimized for reproducibility (separating data, source, configs, and artifacts).
• 🐍 Modern Dependency Management: Automatically initializes Poetry projects, solving dependency hell before it starts.
• ⚙️ Hydra Configuration: Pre-configured Hydra setup for flexible, composable experiment configuration.
• 💾 Smart Presets: Built-in configurations for Computer Vision, NLP, and General ML. Save your own custom setups for future use.
• ✅ CI/CD Ready: Auto-generates GitHub Actions workflows to run your tests on every push.
• 🔧 Git Integration: Initializes a Git repository with a smart .gitignore tailored for ML (ignoring large data/artifacts).
• 📄 Auto-Documentation: Generates README.md files in every subdirectory explaining exactly what should go where.

📦 Installation

You can install mlinit directly from PyPI:

pip install mlinit

Tip: If the mlinit command is not found after installation, run it via python:

python -m mlinit.cli init --name my_project

🚀 Quick Start

Initialize a new project in seconds:

mlinit init --name my_awesome_project

Example Output

The tool generates a ready-to-run train.py integrated with Hydra:

# src/train.py
import hydra
from omegaconf import DictConfig, OmegaConf

@hydra.main(version_base=None, config_path="../conf", config_name="config")
def train(cfg: DictConfig):
    print(OmegaConf.to_yaml(cfg))
    # Your training logic here...

🛣️ Roadmap

• MLflow Integration: Auto-configure MLflow for experiment tracking.
• DVC Setup: Automated data version control hooks.
• Cloud Deployment: Terraform templates for AWS/GCP.
• Streamlit/Gradio: One-click demo app generation.

📂 The "mlinit" Standard Structure

Your generated project will look like this:

my_project/
├── conf/                   # Hydra configuration files
│   ├── base/               # Base configs (train.yaml, model.yaml)
│   └── config.yaml         # Main entry point
├── data/                   # Data directory (gitignored content)
│   ├── raw/                # Immutable raw data
│   ├── processed/          # Canonical data sets for modeling
│   └── external/           # Data from third party sources
├── src/                    # Source code
│   ├── data/               # Scripts to generate data
│   ├── models/             # Model architectures
│   ├── utils/              # Helper functions
│   └── train.py            # Main training script (Hydra-enabled)
├── notebooks/              # Jupyter notebooks for exploration
├── tests/                  # Unit tests
├── artifacts/              # Model checkpoints, logs, outputs (gitignored)
├── .github/workflows/      # CI/CD pipelines (GitHub Actions)
├── pyproject.toml          # Project dependencies (Poetry)
├── README.md               # Project documentation
└── .gitignore              # Git ignore rules

🙏 Acknowledgements

This project was heavily inspired by Cookiecutter and the Cookiecutter Data Science project. We aim to bring that same spirit of community-driven standardization to the modern ML stack (Hydra, Poetry, etc.).

We believe in the power of the open-source community to build better tools together. If you have ideas or feedback, please open an issue!

🤝 Contributing

We welcome contributions! Please see CONTRIBUTING.md for setup instructions.

1. Fork the repository.
2. Create a feature branch.
3. Submit a Pull Request.

---

Built with ❤️ for the ML Community.