aiclimate 0.1.0

A CLI tool that scaffolds Python ML/data science projects for AI in climate and environmental sciences research.

aiclimate

A CLI tool that scaffolds Python ML/data science projects for AI in climate and environmental sciences research.

Built by AI for Climate and Environmental Sciences.

Quick Start

# Install with uv
uv tool install aiclimate

# Scaffold a new project (interactive)
aiclimate

# Or use with an external cookiecutter template
aiclimate gh:user/template

What You Get

When you run aiclimate, you'll be prompted for project details and then a complete project is scaffolded:

my-climate-ai-project/
├── pyproject.toml          # uv-compatible, with ML + climate deps
├── justfile                # Task runner recipes
├── .python-version
├── .gitignore
├── README.md
├── .github/workflows/ci.yml
├── configs/
│   └── default.yaml        # Experiment configuration
├── data/
│   ├── raw/                # Original, immutable data
│   ├── processed/          # Cleaned and transformed data
│   └── external/           # Third-party data sources
├── experiments/            # Experiment outputs and logs
├── models/                 # Trained model checkpoints
├── notebooks/
│   └── 01_exploration.ipynb
├── src/<package_name>/
│   ├── __init__.py
│   ├── config.py           # Configuration loading
│   ├── train.py            # Training entrypoint
│   ├── evaluate.py         # Evaluation entrypoint
│   └── predict.py          # Inference entrypoint
└── tests/
    └── test_placeholder.py

Template Variables

Variable	Description	Default
project_name	Human-readable project name	My Climate AI Project
project_slug	Directory name (auto-generated)	my-climate-ai-project
package_name	Python package name (auto-generated)	my_climate_ai_project
project_description	Short project description	...
author_name	Your name	Your Name
author_email	Your email	...
python_version	Python version	3.12
ml_framework	ML framework (pytorch/tensorflow/jax/none)	pytorch
include_dvc	Include DVC for data versioning	no
license	License type	MIT

Default Dependencies

Projects come pre-configured with common climate/environmental AI packages:

• Data: numpy, pandas, xarray, netCDF4, scipy
• ML: scikit-learn + your choice of pytorch/tensorflow/jax
• Visualization: matplotlib, seaborn
• Utilities: pyyaml, tqdm, loguru
• Dev tools: pytest, ruff, jupyter

Backward Compatibility

aiclimate is built on the cookiecutter engine and can use any existing cookiecutter template:

aiclimate gh:audreyfeldroy/cookiecutter-pypackage

Development

# Clone and install
git clone https://github.com/aiclimate/aiclimate.git
cd aiclimate
uv sync

# Run tests
just test-all

# Lint
just lint-check

License

BSD-3-Clause