Final-year ML student here, trying to break into AI growth. I built a fraud-risk agent that cut expected loss per transaction from ₹936 to ₹88 by tuning the threshold for cost instead of accuracy. Question for growth folks: what's the ML-for-growth problem you wish someone would solve for you?
