MediBot.AI - Medical Chatbot with RAG + Document Analysis
Engineered a RAG based chatbot using LLaMA 3.1 (via Groq API), building a custom vector store of 4800+ text chunks from a 750+ page medical encyclopedia. Built a hybrid retrieval system combining custom BM25 keyword scoring with cosine similarity-based semantic search. Implemented OCR-based document analysis (Tesseract + PyPDF2) for uploaded blood reports and prescriptions, delivering AI-generated summaries with abnormality detection.
// INTERVIEW ENGINE