(Ebook) Kubernetes for MLOps - Scaling Enterprise Machine Learning, Deep Learning, and AI by Sam Charrington

(Ebook) Kubernetes for MLOps - Scaling Enterprise Machine Learning, Deep Learning, and AI by Sam Charrington

Enterprise interest in machine learning and artificial intelligence continues to grow, withorganizations dedicating increasingly large teams and resources to ML/AI projects. Asbusinesses scale their investments, it becomes critical to build repeatable, efficient, andsustainable processes for model development and deployment.The move to drive more consistent and efficient processes in machine learning parallelsefforts towards the same goals in software development. Whereas the latter has come to becalled DevOps, the former is increasingly referred to as MLOps.While DevOps, and likewise MLOps, are principally about practices rather than technology, tothe extent that those practices are focused on automation and repeatability, tools have beenan important contributor to their rise. In particular, the advent of container technologies likeDocker was a significant enabler of DevOps, allowing users to drive increased agility, efficiency,manageability, and scalability in their software development efforts.Containers remain a foundational technology for both DevOps and MLOps. Containers providea core piece of functionality that allow us to run a given piece of code—whether a notebook,an experiment, or a deployed model—anywhere, without the “dependency hell” that plaguesother methods of sharing software. But, additional technology is required to scale containersto support large teams, workloads, or applications. This technology is known as a containerorchestration system, the most popular of which is Kubernetes.