Mr.HelpMate AI — Retrieval Augmented Generative Search (Insurance Policy QA)

A full, experiment-friendly RAG pipeline that answers questions from a Group Member Life Insurance Policy PDF. It includes three layers:

1. Embedding Layer: PDF parsing, text cleaning, and multiple chunking strategies + choice of embedding model (OpenAI or SentenceTransformers).
2. Search Layer: ChromaDB vector store with caching and cross-encoder re-ranking.
3. Generation Layer: Robust prompt + few-shot options with OpenAI Chat (configurable).

It also produces 6 screenshots automatically:

• 3 images showing top-3 retrieved chunks (one per query).
• 3 images showing the final generated answers (one per query).