Machine Learning System Design Interview, issue 3, Nov 25, 2025
The Gradient Drowning Trap
Senior Machine Learning Engineer interview at Google DeepMind, and the interviewer asks:
βWe have a 1:1000 class imbalance for fraud detection. We applied π€ππ’π΄π΄_πΈπ¦πͺπ¨π©π΅π΄ to the ππ«π¨π¬π¬-ππ§ππ«π¨π©π² loss, but the model is still missing the hard edge cases. What do we do?β
The hidden gradient dynamics that SMOTE, class weights, and oversampling canβt fix.
The full answer, with the mechanism and the arithmetic, is for paid subscribers on Substack.