Data/ML/AI
Overfitting
Also written as Underfitting, Generalization
When a model memorises the examples it was trained on instead of learning the general pattern, so it scores brilliantly in testing and fails on real data.
Think of it like
A student who memorised last year's exam paper. Perfect marks on that paper, lost on this year's questions.
Junior or senior?
The most basic ML failure there is. A candidate who cannot describe it plainly has not really trained models.
Senior sounds like
Talks about catching it in practice, not just defining it.
Ask them
“How did you know a model was overfitting rather than genuinely good?”