Lessons From The Field In Building Your MLOps Strategy

Businesses are learning how to drive real value from machine learning. In most cases, this involves assembling an MLOps (machine learning operations) platform to manage and standardize related workflows: data collection and cleaning, feature extraction, experimentation and training, model management, deployment, serving and monitoring. Development of MLOps platforms remains challenging for most teams, due to heterogeneous ML use cases, overlapping and mismatched tooling, and a need to support both experimental R&D and automated CI/CD capabilities. This talk will explore several case studies from industry to illustrate principles and best practices you can apply to building your own MLOps platform.