Where to start? Continuous Delivery for Machine Learning

Face the challenges of real-world ML systems with the MCubed webcast Ep2


As machine learning components make their way into more and more applications, teams are faced with the challenge of transforming the traditional Continuous Delivery process for an environment that often deals with not one but three types of change. Application code, machine learning models, and training data all evolve and need to be carefully considered when coming up with a catch-all workflow.

That’s not everything, though - different divisions need quite different tools to build the artifacts that make up the final ML software product and of course they all need some form of integration. Plus there’s always the question of reproducibility should something go wrong. Let’s be honest, it all can feel pretty daunting on first looks.

Luckily, you’re not the first person who has to try and get a grip on this complexity, and we’ll have some help available in this webcast. In the second installment of our MCubed webcast series, automation expert Danilo Sato will walk you through the components of a good continuous delivery system and help you understand what is necessary to provide users with a stable, current ML application that makes use of all the latest assets.