The thread for this Webinar is here.
If you missed the presentation see the slides and recording here.

Abstract: 

Vector search alone quietly fails on certain kinds of queries, like exact error codes, product IDs, and rare terms, while keyword search misses results phrased differently from the query. Hybrid search solves this by running lexical and dense vector search over the same data and fusing the results, so a confident hit from either retriever still ranks well. It has become the standard retrieval approach for RAG, and increasingly for enterprise search. If your data already lives in Db2, you already have the building blocks for hybrid search over your own data. This session shows how to build it end to end inside Db2 12.1.5, from ingestion to query: one chunk table, two representations, two indexes, and reciprocal rank fusion expressed in SQL. You'll leave understanding the full ingestion-to-query pipeline, including why a strong single-retriever hit is never penalized. Aimed at engineers and architects on the Db2 platform who want AI-powered retrieval without standing up a separate vector database.

Speaker(s):  
Shaikh Quader

Shaikh Quader has spent over two decades at IBM in software development, with the last ten years focused on AI. Now serving as AI Architect and Master Inventor on IBM Db2, he works on building intelligent, high-performance data platforms — including capabilities such as vector search and LLM integration. He holds many issued patents in AI and data systems and has published peer-reviewed research on machine learning. He is also a Ph.D. candidate at York University, where his research explores the intersection of AI and relational databases.

David Kalmuk

David Kalmuk is a Distinguished Engineer, Master Inventor and Chief Architect for Databases at IBM. He leads the technical strategy for Databases within Core Software Products which spans the Db2 family of products including Db2, Db2 Warehouse, Db2 SaaS, Db2 pureScale, as well as Netezza.