Background
Kaira is an open-source toolkit used in the study of communication systems, allowing researchers to design and develop methods to ensure effective transmission of messages from sender to receiver. Written in Python and built on the PyTorch machine learning ecosystem, it is designed in a modular way to allow easy extension and integration with existing PyTorch projects, enabling rapid prototyping of existing and novel communications systems.
Our Contribution
We were contacted by the researcher as part of the Open Source Booster programme, with the goals of improving documentation, updating developer infrastructure, and boosting the project's impact. By conducting an audit of the repository, we identified several areas for improvement, particularly around developer documentation and infrastructure, including suggestions for better use of pre-commit hooks, and improvements to release notes and the contributor guide. We also provided suggestions for publicising the project to reach a wider audience, as well as instructions for making the project citeable, such as registering on Zenodo.
Outcomes
It is hoped that, once implemented, these suggestions will help reduce friction for new developers to get involved with the project, and help new users to discover, use and cite the tool.