NUTRICOACH : A MULTI-AGENT AI FRAMEWORK FOR PERSONALIZED NUTRITION RECOMMENDATION USING RAG AND LLMS

Multi-Agent Artificial Intelligence systems are increasingly employed for complex, domain-specific tasks requiring collaboration and contextual reasoning.This paper presents NutriCoach, a multi-agent AI framework for personalized nutritional analysis built using React, FastAPI, Groq LLaMA 3.1, and Astra DB.The system comprises four specialized agents -a Calorie Estimation Agent that infers macronutrients from natural language food descriptions, a Diet Recommendation Agent that computes personalized meal plans using the Mifflin-St Jeor formula adjusted for user goals and dietary restrictions, an Ask-AI Coach Agent that delivers evidence-grounded responses using the user's health profile and food history, and a RAG-based Knowledge Retrieval Agent using Sentence Transformers with cosine similarity search over a structured nutrition knowledge base.Per-user authentication is implemented via JWT tokens and bcrypt