At InstaDeep, I develop deep learning models for computational pathology, including segmentation pipelines for gigapixel whole-slide images and models that classify cell types. I contribute across preprocessing, training, validation, and evaluation.
I also optimized production inference by porting models to JAX and using XLA compilation to reduce prediction times. For therapeutic peptide design, I re-implemented a high-compute training pipeline in JAX and trained sequence-based GANs using the LeakGAN architecture.
My work spans computer vision, computational pathology, and generative AI for bioinformatics. I hold an engineering degree in ICT Engineering from Higher School of Communication of Tunis (SUP'COM).
Poland