GPT-5.5: Vector Database RAG Optimization

Explain how to optimize a Retrieval-Augmented Generation (RAG) system using a vector database like Pinecone or Milvus. Discuss strategies for chunking strategies (fixed-size vs. semantic), embedding model selection, and re-ranking techniques to improve the relevance of retrieved context.