Embedding a big dataset of descriptions

Hey!
We have a 800k rows of item descriptions which we want to embed with ada2 - any idea how to do this not to recalculate it on each deployment (incremental?)

Yes, incremental builds should help you here.