How Airbnb evaluates coding models
How we built a benchmark from our engineers' work, stopped models from grading themselves, and now run it on every release.
CASTLE: training natural language search before the first real query
How we addressed the cold-start problem for Airbnb’s natural language search by jointly generating synthetic training data — queries and relevance labels — from de-identified booking and search signals.
Personalization without user identity
How Airbnb uses proximity signals to personalize without relying on individual user history.