On REACTA, my graduation project, I developed an R-GCN model to predict drug-drug interactions on a biomedical knowledge graph, achieving 95% Test F1-Macro. I combined molecular fingerprints, BiomedBERT embeddings, and ATC classifications in the model's drug features.
Through the Digital Egypt Pioneers Program (DEPI), I developed a real-time Driver Monitoring System using MobileNetV2 for drowsiness detection and YOLO for distraction detection. I integrated visual and audio alerts, with optional event logging to CSV.
I also created a Smart Career Guidance System that uses NLP to match student CVs with job descriptions. Its Streamlit app achieved 99% accuracy in career recommendations and enabled 200+ students to explore career paths.
