Project goal
This sample project demonstrates the evolution from document retrieval to answers grounded in retrieved content.
Three stages
- Keyword search retrieves documents using indexed terms.
- Vector search retrieves chunks using embeddings.
- RAG passes retrieved context to a language model to produce an answer with source references.
Infrastructure decisions
The frontend stays static while backend APIs coordinate retrieval and model calls. Chunk metadata retains source references so the user can inspect the supporting document.
Verification to document
Compare the same questions across keyword and vector retrieval. Check whether generated answers are supported by the retrieved context, and record latency and cost. Replace these suggested checks with your own evidence.
Platform evolution (Mermaid sample)
flowchart LR
A[Keyword search] --> B[Vector retrieval]
B --> C[RAG with citations]