All Vector Search is Hybrid Search

This talk starts with a strongly-stated central premise: all vector search applications benefit from some form of hybrid search. This can range from a traditional keyword-based search to metadata filtering and more. Inevitably, vector search applications become search applications, and so need to leverage and reconcile other kinds of search scoring functions as well as more sophisticated search modalities like two-stage retrieval and reranking. Further, hybrid search systems must now contend with the very difficult problems of real-time updates to both vectors and vector metadata, and the difficulties of updating these vector indices to keep data fresh. Join this talk and learn about the questions you should be asking yourself as you embark on your hybrid search infra journey.

John Solitario

Rockset

John Solitario is a senior product manager at Rockset and machine learning and ingest product development at Rockset, which includes vector search, integration support, ingest transformations and collection management. Before joining Rockset, John spent several years at Salesforce where he worked across a multitude of products including Mulesoft, Commerce Cloud, and Service Cloud. He received a B.S in engineering and an M.S. in computer science with a focus in artificial intelligence from Stanford University.