Your question is Extreme Imbalance in Binary Classification. Take a moment with it on the right.
Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).
You are training a deep neural network for a binary classification task where the positive class is very rare. You need a training setup that can still learn useful signal from the minority class without collapsing to majority-class predictions.
How would you handle extreme class imbalance when training a deep neural network for binary classification, both at the data level and the loss-function level?