in-memory index only?

[mindplay-dk]

it's an in-memory index only?

this probably should be clarified in the README - you might want to discuss the memory characteristics as well, like what can be expected in terms of memory usage per document etc. (esp since you mention this is designed to run on edge servers, which typically have memory limitations.)

any plans to support a disk based index for larger volumes of vectors?