substra

substra is a software platform designed to facilitate the development and deployment of machine learning models, particularly within enterprise environments. It aims to address the challenges of reproducibility, collaboration, and governance in the machine learning lifecycle. The platform provides tools for data management, model training, experimentation tracking, and model deployment. One of its core features is the concept of a "data scientist workbench" which offers a consistent environment for data scientists to work in, regardless of their underlying hardware or operating system.

substra emphasizes the importance of secure and auditable data usage, allowing organizations to manage access permissions and track how data is used for training models. It supports various machine learning frameworks and libraries, aiming for flexibility in adoption. The platform also includes features for model versioning and comparison, enabling users to track the evolution of their models and understand performance differences. Collaboration is a key aspect, with functionalities designed to allow multiple users to work on projects together, share code, and manage shared assets. Ultimately, substra seeks to streamline the operationalization of machine learning by providing a structured and controlled environment for the entire model development and deployment process.