I am building a RAG with llamaindex and

The community member is building a Retrieval Augmented Generation (RAG) system using llamaindex and wants to connect multiple modules, including a SQL query engine, vector search, keyword search, and image search (OpenAI CLIP). They are looking for guidance on how to build such a system with more than two retrievers. The comments suggest using a combination of different query engines handled with RouterQueryEngine or LLMMultiSelector to do all the searches simultaneously and then pass the results to a cohere reranker to get the final top_k results. There is no explicitly marked answer in the comments.