Question

A Generative AI Engineer is building a RAG application on Databricks. They created a self-managed Delta Sync Vector Search index and populated an embeddings column by calling a model serving endpoint that produces 1024-dimensional vectors. At query time, the retrieval step consistently returns errors or nonsensical results. The engineer confirms the index was created successfully and documents were synced. What is the MOST likely root cause of the poor retrieval behavior?

Accepted Answer

The query embeddings are being generated with a different embedding model than the one used to populate the index, so the vector spaces are incompatible.

Answer

Delta Sync indexes cannot be used with self-managed embeddings and must always use Databricks-computed embeddings.

Answer

The primary key column was not set as a BIGINT, which breaks similarity scoring.

Answer

Vector Search only supports embeddings up to 768 dimensions, so 1024-dimensional vectors are truncated.