MediQuery

This freelance project focused on developing an AI-driven medical chatbot designed to deliver accurate and contextually aware answers to medical questions. The system was built using Llama2 with a Retrieval-Augmented Generation (RAG) approach for enhanced knowledge grounding.

I implemented a document ingestion pipeline and semantic search using FAISS to efficiently retrieve relevant information from large datasets and PDF documents. The conversational layer was developed using Chainlit, offering an intuitive chat experience optimized for real-time medical Q&A.

The model was optimized through quantization techniques to ensure lightweight CPU-based inference and scalable deployment. The architecture also supports continuous knowledge expansion, allowing adaptation to new medical fields or datasets with minimal retraining.

• Developed an AI-driven medical chatbot using Llama2 with Retrieval-Augmented Generation (RAG).
• Implemented semantic search and document retrieval using FAISS and LangChain.
• Built a conversational interface with Chainlit for real-time interactions.
• Enabled PDF ingestion and processing for dynamic medical knowledge retrieval.
• Optimized model performance via quantization for CPU-friendly deployment.
• Designed for scalability and adaptability to multiple medical domains.

Chat Interface

Built a conversational interface using Chainlit, providing an intuitive and real-time chat experience for medical queries.

RAG Pipeline

Implemented a Retrieval-Augmented Generation pipeline combining document ingestion, semantic search (FAISS), and Llama2-based response generation.

Knowledge Base System

Designed a dynamic document ingestion and PDF processing system enabling continuous expansion of medical knowledge and adaptability to new domains.