In a white paper titled “Practitioners Guide to MLOps,” Google experts Khalid Salama, Jarek Kazmierczak, and Donna Schut discuss the concept of MLOps and its lifecycle. The paper raises questions to help evaluate an organization’s current ML strategy. The questions are grouped into different stages of an ML delivery pipeline and cover automation, collaboration, reproducibility, governance, and compliance. The questions focus on data acquisition and exploration, data transformation and feature engineering, experiments, model training, and evaluation, deployment and serving, and model pipeline, monitoring, and continuous improvement. A MLOps system is recommended to streamline and structure the ML strategy. Future blogs will demonstrate the implementation of MLOps best practices using different technologies.

source update: 31 Questions that Shape Fortune 500 ML Strategy – Towards AI