The MLOps movement and associated new tooling is starting to help tackle the very real technical debt problems associated with machine learning in production.
MLOps isn't just about new technologies and coding practices. Getting better at productionizing your models also likely requires some institutional and/or organisational shifts.