Abstract: Vector databases are transforming the way we interact with our data. You can search across different data sources (text, images, audio, etc.) and store them as a vector embedding. Vector embeddings are just an array of numbers that represent the data object. For example, [-0.12, 0.02, …, 0.22] can represent an image of a dog in our database. To enable this, we use machine learning models to transform our data and capture its semantics in the vector. Scaling ML models to work reliably in production is complex, and implementing efficient vector search while offering CRUD support is even harder.
Vector databases are an elegant solution to these challenges. It searches through your data quickly with Approximate Nearest Neighbor (ANN) search while supporting full CRUD operations. Weaviate is an open-source vector database that enables developers to build various search and recommendation applications.
In this session, you will learn more about vector embeddings, how vector search engines work, why they are so fast, and how they could help you take your search to production. I will also share a live-coding demo, showing you all the steps from setup to query execution.
Bio: Erika Cardenas is a Developer Advocate at Weaviate, an open-source vector database. She has two master's degrees in economics and data science from Florida Atlantic University. Erika was part of the NSF-NRT program, where she published a paper on predicting house prices using structured and unstructured text data. She has written several blog posts such as vector database versus vector library, hybrid search, and integrating LangChain and Weaviate for generative search applications.