DevOps for AI – Challenges in Development of AI-enabled Applications
Lucy Ellen Lwakatare, Ivica Crnković, Jan Bosch
Abstract
When developing software systems that contain Machine Learning (ML) based components, the development process become significantly more complex. The central part of the ML process is training iterations to find the best possible prediction model. Modern software development processes, such as DevOps, have widely been adopted and typically emphasise frequent development iterations and continuous delivery of software changes. Despite the ability of modern approaches in solving some of the problems faced when building ML-based software systems, there are no established procedures on how to combine them with processes in ML workflow in practice today. This paper points out the challenges in development of complex systems that include ML components, and discuss possible solutions driven by the combination of DevOps and ML workflow processes. Industrial cases are presented to illustrate these challenges and the possible solutions.