Does anyone know if it's possible to make this work on RDS (e.g. by getting Amazon to install it for you or something?) Being able to store sentence embeddings in Postgres and compute similarity on-the-fly would save a lot of work!