Good ML pipelines ensure reproducibility of ML experiments and controllability of the development process. In practice, there are often situations when you want to reuse the code of one project into a new one. Sometimes, a new project (model) differs only in the target variable. In such cases, you can reuse up to 95% of the developments from the previous project. This talk discusses the approaches to organize and configure ML pipelines using DVC, ways to reuse ML pipelines, and typical scenarios where this can come in handy.
Rozhkov Mikhail - Solution Engineer at Iterative.ai. ML Engineer and enthusiast with over six years of experience in Machine Learning and Data Science. Co-creator ML REPA, author of courses on automating ML experiments with DVC and MLOps. As a member of the Iterative.ai team, he helps teams improve ML development and automate MLOps processes.
This post is for subscribers onlySubscribe
Already have an account? Log in