• Rewrote legacy abuse-detection service within distributed system to address more varied types of spam (Java 8, MySQL, Redis).

• Developed rule-based system to identify abuse based on social actions within the mobile and web applications (e.g. commenting, following, etc).

• Yielded 93% more alerts within a one month timespan, with an initial 14% false-positive rate which was reduced to 4% after reconfiguration of rule-based system.