In this session, you’ll learn how Airbnb is leveraging OpenSearch as a vector database for its embedding platform. This talk is for everyone, starting with an overview of how Airbnb is using vector embeddings to solve guest, host & marketplace challenges, followed by specific use cases that led Airbnb to adopt OpenSearch as its vector database. We will share lessons learned and provide insights into strategies for optimizing OpenSearch as a vector database.This talk is must attend for those exploring vector databases to power embeddings based retrieval, generative AI or RAG use cases.
In this talk we will
- Overview - Discuss how we use embeddings and LLMs to tackle many product challenges across Airbnb.
- Infrastructure - We will also discuss the infrastructure we built at airbnb to power vector solutions with focus on Airbnb’s embedding platform.
- OpenSearch - This section will discuss in detail how we leverage OpenSearch as a vector database to store and retrieve embeddings effectively.
- Optimizations - Finally we share our learnings and strategies to optimize OpenSearch vector indices with “image search” as case study.
