Watch the event replay to learn how AI is transforming our Relevance and Customer Support teams. We also dive into Chronon, Airbnb’s Open Source data platform for serving for AI/ML applications.
AI for Infinite Choice
Sanjeev Katariya | Director, Engineering – Relevance and Personalization
In a world of infinite choices, Airbnb is reimagining how personalized discovery works at global scale. This talk explores how ranking, recommendations, retrieval, and GenAI come together across Airbnb’s platform to guide guests and hosts through diverse, dynamic journeys. From large-scale learning-to-rank systems and embedding-based search to contextual bandits, generative metadata, and agentic AI, we’ll share how intelligence is woven into every touchpoint—along with a look at what’s next in adaptive, modular AI for discovery.
How to Answer Millions of Guest & Host Questions in Seconds
Yashar Mehdad | Principal Engineer, Community Support Engineering
Airbnb’s mission is to deliver effortless, world-class support for every guest and host. This talk unpacks the Airbnb AI Assistant that makes that possible—an end-to-end system that fuses rich and high quality data, robust evaluation platform, models, and guardrails. We’ll discuss how these components are designed to interact with the goal of delivering responses that are fast, accurate, and aligned with applicable policies, while incorporating feedback to support ongoing improvement. Attendees will gain insights into approaches and considerations for designing, deploying, and scaling robust AI assistants intended for community support.
Chronon: Democratizing High-Performance AI/ML Feature Engineering Through Open Source
Pengyu Hou | Senior Software Engineer, Infrastructure
Chronon, Airbnb’s innovative declare feature platform, is designed to address the unique challenges faced by AI/ML practitioners.In this talk, learn about Chronon’s architecture that guarantees online-offline consistency through a shared computation engine. This session covers high throughput low-latency feature serving, Python API functionality, feature derivation capabilities, and recent enhancements. We’ll also explore future roadmap items including potential integration with unstructured data.