Observability in DevOps?

The problems that we're solving with MLOps are actually a super set of the problems in DevOps. Machine learning models are also, I don't want to say non-deterministic,. But they have a greater degree of dynamism in them because they're fundamentally relying on processes from the real world That requires tracking metrics like, is my input data drifting? You know, is my F1 score going down? All of these different things.