Design Personalized Promotion Recommendations
Company: Creditkarma
Role: Machine Learning Engineer
Category: ML System Design
Difficulty: medium
Interview Round: Technical Screen
Design an offline personalized recommendation system for promotional products. The system should periodically generate ranked product or offer recommendations for each user, to be served later through channels such as email, push notifications, or an in-app surface.
Pay special attention to the fact that the main business outcome is conversion, but conversion feedback is delayed and sparse. Explain how you would design the major components, including data collection, feature engineering, candidate generation, ranking, feedback labeling, offline training, evaluation, serving, monitoring, and experimentation.
Quick Answer: This question evaluates a candidate's ability to design end-to-end machine learning recommendation systems, covering competencies such as data collection and pipelines, feature engineering, candidate generation, ranking, delayed and sparse conversion feedback labeling, offline training and evaluation, serving, monitoring, and experimentation.