The model is fixed and the world is not, and an MLOps interview is about noticing when the two have parted: which run produced this number, why the feature differs in production, why the pipeline is green and the dashboard is wrong.
These are the skills in our catalogue that this role draws on. They are not the whole interview.