Case Study: Moasis Achieves AI-Powered CTR Prediction and Bid Optimization with KUNGFU.AI

Moasis - Customer Case Study

Moasis, an ad tech firm, wanted to optimize client ad spend by predicting click-through rates (CTR) from bid request data such as location, time, app, and device signals. After a prior vendor engagement fell short, Moasis turned to KUNGFU.AI to evaluate whether machine learning could work with their “spotty” real-time bidding data and strict latency constraints.

KUNGFU.AI built a CTR prediction and bid price optimization solution using a Gradient Boosted Tree model with lightweight bid logic, deployed as a Dockerized microservice on AWS EC2. The system achieved sub-10 millisecond inference, processed more than 3,000 bid requests per second, reached an AUC of 0.805, and drove a 10X improvement in A/B testing, giving Moasis a proprietary ML-powered differentiator.

Moasis