A proactive security framework for multi-cloud Identity Access Management (IAM). This project implements semantic policy analysis and risk detection using quantized Large Language Models (LLMs) to eliminate misconfigurations without exposing sensitive data.