MLOps is what turns a model that works in a notebook into a model that works in production and keeps working. It covers versioning data and models, automating training and deployment, monitoring for drift as the real world changes, and being able to roll back to a known-good version when something degrades.
The reason it exists as a discipline is that models fail differently from ordinary software: they can be running perfectly and still becoming wrong, because the data flowing through them has shifted. Catching that requires monitoring the model’s behaviour, not just its uptime.
Related terms