Research would like to start working with OpenSearch, to experiment with vector embeddings. This would be a good opportunity for Search and Research get hands on experience with this kind of search.
I would suggest the following (sub) tasks:
- create an instance/a cluster
- set up an index that resembles one of our content indices
- vector creation:
- [Option A] configure an ingestion pipeline to create embeddings server-side
- [Option B] calculate the vectors client-side (jupyter?)