What you'll learn

As the world's top-grossing mobile app, payment processing is a crucial part of Tinder. Scaling the payment system is very challenging with a legacy, untestable code base. To resolve the problem, Tinder's team recently made the decision to build a state-machine-based, pluggable solution to represent a universal purchase flow, which is agnostic of any specific payment methods. In this talk, Yisong Wu and Fernando Pereira will share how Tinder's state machine / DSL approach ensures the flow is deterministic, self-documented, and easily testable and how its plug-in system guarantees flexibility by allowing dynamic custom rules to be run along the state transitions.