Machine Learning Operations (MLOps) is a set of technologies and best practices that streamline the management, development, deployment, and monitoring of data science models at scale across a diverse enterprise. MLOps is multidisciplinary, often bringing together data scientists, DevOps engineers, ML Architects and software developers.
MLOps principles and platforms provide organizations with repeatability and consistency within the data science discipline, allowing for teams to:
While both DevOps and MLOps involve code, pipelines, and continuous cross-functional communication, the behavior of software in production is predetermined whereas models' behavior changes over time. MLOps takes DevOps development standardization principles into account within the context of machine learning models. Data Science is experimental, with many rapidly more rapidly changing parts compared to software development. As a result, there are more complexities to keep track of that MLOps focuses on specifically, in addition to the probabilistic behavior of models.
For organizations that want to efficiently scale their data science operations, they not only need to support enterprise-grade development and production, but they also need to follow the same standard process that data scientists use.