Speaker
Abstract
Trust and safety challenges become exponentially harder at marketplace scale, especially when interactions happen in real time. In this talk, I’ll walk through how we built SafeChat, DoorDash’s AI-powered moderation system that proactively protects Dashers and customers during in-app interactions.
I’ll cover the evolution of our moderation architecture, how we balance speed, cost, and accuracy using layered ML systems, and what it takes to operationalize AI safely in production. The talk will focus on real engineering and ML trade-offs, lessons learned, and how to design safety systems that are both effective and responsible.
More from the program
From Hype to Habit: Conquering AI Adoption at Zoox
Jatin Aneja · Zoox
From Fab To Token - The State Of The Market
Jordan Nanos · SemiAnalysis
Beyond the Prototype: Scaling Framework Agnostic AI Agent Infrastructure with Ray
Deepak Mohanakumar Chandramouli · Apple, Bhumik Vinodkumar Thakkar · Apple
See all sessions ›