Vector Podcast with Louis Brandy, VP Engineering, Rockset
Second episode of Season 3: I’ve had a pleasure to talk to and learn from Louis Brandy, who is leading the engineering at Rockset, search and analytics database.
In my blog post on vector databases I focused on then emerging class of databases, that treated vectors as first-class citizens. Some of them were also established search engines, but overall all of them were and continue to be their own class.
But lately, a new class of players began to embark on the journey to implement vector search support. One of them is Rockset — SQL database, that can scale massively and gives you the best of both worlds: SQL and vector search.
This means, that you can run your analytics workloads together with applying cutting-edge AI techniques, such as finding nearest neighbors to your query using an LLM. Rockset builds an AI layer atop the RocksDB, and recently launched hybrid search support.
Louis goes to quite a bit of length to explain very interesting details of hybrid search, which does not reduce to only having a linear interpolation between non-vector and vector search retrieval. Take a look!
RSS + your favorite audio platforms: https://rss.com/podcasts/vector-podcast/1460893/
