At ONCF Group, I designed an end-to-end AI-powered visual inspection system to detect residue on phosphate wagons. I developed a real-time video analysis pipeline and integrated it into an application for configuration, monitoring, results analysis, and KPI tracking.
At Moroccan High-Speed Train Maintenance Company, I developed a Computer Vision and OCR solution to identify railway components. I also implemented workflows for collecting, preparing, annotating, and managing data, including tools to enrich the dataset with new component references.
On my Embedded Driving Assistance System project, I designed and deployed a low-cost, real-time system for Moroccan traffic-sign recognition, integrating a multilingual voice assistant. I also developed a RAG-based assistant that brings hospital information together for natural-language questions.
My projects include training a 3D U-Net for brain tumor segmentation, which achieved a 0.93 Dice score for the whole-tumor region, and developing a multimodal recommendation system that achieved a 0.977 AUC on offline validation.
