Retrieval-Augmented Research
A live-document RAG pipeline comparing dense, MMR and intent-enriched retrieval.
KEY SIGNALThree retrieval variants
THE CHALLENGE
What needed to change.
Live sources change, retrieval can become redundant and user questions may not match document vocabulary.
THE APPROACH
How I structured the solution.
I created dense, dense-plus-MMR and intent-enriched retrieval variants with Llama 2, LangChain and ChromaDB.
THE OUTCOME
What the work demonstrated.
The project produced a structured basis for analysing how retrieval diversity and query enrichment affect answer support.