🩺 Specialist Search Agent (SSA) – RAG Pipeline

This project explores a multi-agent, Retrieval-Augmented Generation (RAG)-powered AI system designed to intelligently match patients with medical specialists based on symptoms, location, insurance, and availability.

Developed as part of a technical assignment for the One-Click Referral System, this system parses natural language patient queries and returns a ranked list of specialists with clear, explainable justifications.

---

🚀 Features

• Natural Language Understanding using zero-shot classification (facebook/bart-large-mnli)
• Semantic Search over specialist profiles using vector embeddings (HuggingFace models)
• RAG Pipelines with FAISS, LangChain, and LlamaIndex + Chroma
• Specialist Ranking based on:
  • Symptom-specialty alignment
  • Insurance compatibility
  • Geographic proximity
  • Next-3-day availability
• Patient-Friendly Justifications via Falcon-7B (tiiuae/falcon-7b-instruct)
• Multi-Agent Architecture:
  Specialist Search Agent → Supervisor Agent → Insurance Agent
• Reflection Loop: Auto re-query on low-confidence results
• API-ready Placeholder: Simulated logic for real-time availability updates

---

📂 Project Structure

notebooks/
├── Multiagent_RAG_pipeline_final.ipynb  # End-to-end pipeline
data/
├── Mock_Specialist_Dataset.csv          # Metadata for doctors
doc/
├── Shreya Banik-Interview with Sri.pdf         # Task document