Example project showing how to use hypvector for in-browser vector search against a parquet file in cloud storage.

The demo embeds search queries in the browser with transformers.js (Xenova/all-MiniLM-L6-v2, 384 dimensions) and runs searchVectors directly against a 249 MB parquet on S3 — no backend, no vector database. Each query reads ~6 MB across ~160 ranged HTTP fetches.

npm i
npm run build

The build artifacts will be stored in the dist/ directory and can be served using any static server, eg. http-server:

npm i -g http-server
http-server dist/

• hypvector is currently consumed via a local file: dependency (../../hypvector). Once it is published to npm this can be switched to a version range.
• The vector parquet on S3 (s3.hyperparam.app/hypvector/wiki_en.vectors.parquet) was built by running npm run data:embed in the hypvector repo against the 156k English Wikipedia sample and uploaded with aws s3 cp ... --profile hyperparam-platypii.