AI in Scam Detection: Coursework with LLMs

Created using ChatSlide

This coursework outlines the development of 'Scam Sentinel,' an AI and cybersecurity project focused on training LLMs to detect fraud patterns in real-time across multiple channels. Designed by Bangalore Institute of Technology, the project emphasizes scalability, explainable predictions, and effective multi-channel fraud detection using models like Llama 3.1 and Mistral 7B, deployed on Google Colab. The methodology includes QLoRA fine-tuning, synthetic dataset expansion, and token-level...