The Question
ML DesignAdversarial Spam Detection at Scale
How would you design a robust, real-time spam detection system for a global social media platform that handles billions of posts daily? The system must minimize the visibility of malicious content while maintaining extremely low false-positive rates, and it must be capable of adapting to rapidly changing adversarial tactics used by spammers.
Naive Bayes
XGBoost
BERT
GNN
Real-time Features
February 25, 2026