Almost Everything can be Vector Search

Roel M. Hogervorst


Categories: blog Tags: 100DaysToOffload podcastepisode

I listened to Data engineering podcast -‘real time vector search’

The interviewee is a vendor, but nevertheless the talk about vector databases was interesting and not a pitch for rocksetdb.

They talked about vectors, as they are used in ML (machine learning), for instance in embeddings and large language models. Vectors are a numerical representation of something, with the idea that they live in a more semantic space. points in that space that are closer together are semantically more similar. But technically it is a an fixed size array for every row in your table.

Some ideas I wanted to highlight:

I’m publishing this as part of 100 Days To Offload. You can join in yourself by visiting, post - 16/100

Find other posts tagged #100DaysToOffload here