When AI Leaves the Lab: Deep Learning's Practical Breakthroughs in IoT and Medical Imaging

This research addresses a fundamental challenge in applied AI: most academic papers validate models in controlled labs but never prove they work in the real world. The undergraduate demonstrates that deep learning models can generalize across different datasets for IoT intrusion detection and run efficiently on edge devices—removing key barriers to real-world security deployment. In medical imaging, the advancement to 8-class cancer grading with explainable AI (GradCAM) shows how to make AI clinically actionable, not just statistically accurate.