2026-05-18T01:56:23Z https://yabu.me Marcos Huerta https://yabu.me/npub1x6eq72x7a3my3gs4mkyejatderq7w6xm9qxz23aql9ey25xw89cqc5pekt https://cdn.vmst.io/accounts/avatars/109/430/021/953/230/080/original/91179ead3dd31e4a.jpeg https://cdn.vmst.io/accounts/avatars/109/430/021/953/230/080/original/91179ead3dd31e4a.jpeg https://yabu.me/nevent1qqs9u2y0wj7anzunjjw2663sm7csyk6h2cac92cst9mtrngqxxe2dhgzyqmtyregmmk8vj9zzhwcnxt4dhyvremgmv5qcf285ruhy32secuhquysemu In reply to <a href='/nevent1qqsgs48ul2nsgv4m9a5khg0y0ac6tmua2shxuknz2kv4xwkkwdjuekq0ykxlc'>nevent1q…kxlc</a><br/>_________________________<br/><br/>Very cool! I gave a talk at Posit Conf about low-memory semantic search using DuckDB for the vector DB (and LLamaCPP for live encoding):<br/><br/><a href="https://www.youtube.com/watch?v=casO6fLXxLw">https://www.youtube.com/watch?v=casO6fLXxLw</a><br/><br/>It works well enough on small text and - most importantly - runs on my 2GB of RAM droplet. <br/><br/>Talk notes (links, example code, etc) here:<br/><br/><a href="https://marcoshuerta.com/posts/positconf2025/">https://marcoshuerta.com/posts/positconf2025/</a> 2026-04-05T19:06:11Z