Situated at the intersection of R&D and IT operations, MLOps is crucial to realising the potential of Data Science: after all, what good is a model that’s superbly accurate but never used? Why capture exquisitely complex relationships automatically yet require constant human supervision and intervention? During this talk, we will examine some common hurdles to operationalisation across the entire Data Science lifecycle, from data engineering and modelling through to deployment and monitoring. We will highlight similarities and differences with traditional DevOps best practices and discuss their applicability in Data Science. We will also discuss some implementation details using both open-source and commercial solutions.