Continuous Delivery for Machine Learning: Patterns and Pains

Implementing Continuous Delivery in a software context is already a challenge, but when it comes to building data-driven systems with machine learning, complex data architectures, and evolving demands, achieving a true Continuous Delivery state is even more difficult. From a lens of extensive experience in developing continuous delivery practices in software contexts, this talk will explore where things stand in achieving CD for data systems in practice, look at common pain points, and explore the patterns we regularly see in addressing these challenges.