Feature Request: Support for Maximal Marginal Relevance (MMR) in Vectorize

Hello,

I’d like to request support for Maximal Marginal Relevance (MMR) in Vectorize.

MMR is a well-known technique for improving result diversity while maintaining relevance in vector search. It helps reduce redundancy in retrieved results by balancing similarity with novelty. For reference, see:
https://www.cs.cmu.edu/~jgc/publication/The_Use_MMR_Diversity_Based_LTMIR_1998.pdf

I’ve reviewed the current documentation and API reference but couldn’t find any indication that MMR is supported. If it already exists, I’d appreciate guidance on how to use it.

If it’s not currently supported, I believe this would be a valuable addition to Vectorize. Since MMR operates as a post-retrieval re-ranking step, it could potentially be implemented as an optional relevance/diversity filtering layer on top of existing similarity search results.

Proposed capability:
- Optional MMR-based re-ranking for query results
- Configurable trade-off parameter (e.g., lambda) between relevance and diversity
- Seamless integration with existing query APIs

This would significantly improve use cases like RAG pipelines, search result diversification, and recommendation systems.

The issue names no files, tests, or entry points. Start by locating Vectorize’s existing query API and determine how optional MMR re-ranking and a configurable lambda would fit; done means the proposed capability is defined, integrated with queries, and covered by appropriate tests.

Written by the indexing model from the issue text.