Rezkellah Noureddine Khiati

• • Designed and trained deep learning models (diffusion, GAN, graph networks, 3-D segmentation) on clinical CT data across multiple pathologies.

• • Managed full experimental pipelines: data curation (DICOM→NIfTI), distributed multi-GPU training, reproducible evaluation with W&B.

• • Published 5 papers at top medical imaging venues. Wrote quarterly technical reports for industry partner.

• • Benchmarked segmentation architectures (U-Net, TransUNet, Swin-UNETR) with systematic ablation studies.

• • Deployed pulmonary disease detection models on clinical data with radiologist-aligned evaluation.

• • Built joint visual–contextual embedding systems (CNN + Transformer) for large-scale image retrieval.