Training ML models can be time-consuming, but the real challenge lies in integrating an ML system into a production environment – in other words, into the software product that users interact with.
An ML system is comprised of three main elements: the training data, the ML model, and the code used to train the models. We apply DevOps principles to
ML systems (MLOps) to combine the development and operations of ML. MLOps, as an extension of DevOps, focuses on automating and monitoring every step of
integrating ML systems into software projects.
In this primer, we explain the fundamentals and principles of MLOps, aiming to provide insights into MLOps processes from an engineering perspective.
innoq.