Time Slot: Track 1 12:20-12:50
Speaker: Mr. Peter Ho | Red Hat | Hong Kong
While Python’s application in developing machine learning models is currently one of the most popular subjects, transforming a machine learning prototype into a production-grade system is complicated. Many people tends to concentrate on the machine learning algorithm itself, while other critical technical aspects, such as automated training pipelines, model performance tracking, model versioning, and model deployment, often get overlooked when trying to transition a machine learning model into production. This has led to numerous machine learning ideas not making it past the concept stage and into practical, real-world use.
MLOps is a paradigm to address these challenges. In this introductory section, we will talk about the very basics of MLOps. We’ll also introduce Kubeflow, a Kubernetes-based MLOps platform that puts MLOps into practice. Furthermore, we’ll explain the fundamentals of the Kubeflow Pipeline Python SDK, demonstrating how Python and Kubeflow can work together to help you get started with MLOps.
Mr. Peter Ho
Peter Ho is one of the Specialist Solution Architects at Red Hat Hong Kong. He has extensive experience in working with Kubernetes, web development and public cloud technologies, with certificates received from Red Hat, Linux Foundations, Microsoft Azure and Amazon Web Services. Peter is passionate about new technologies, and enthusiastic to share any knowledge that he has gained from his application development and public cloud adoption projects.