Yasmine Chaiben

- Contributing to the European Pods4Rail project by developing AI-driven positioning solutions for safe and reliable transportation systems. - Designing and evaluating machine learning models for real-time environmental context detection using multimodal data from fisheye images and GNSS signals. - Developing predictive models for positioning error estimation and correction, validated on real-world trajectories to improve localization accuracy and system robustness. - Building end-to-end ML workflows, including data preparation, feature engineering, model validation, and integration into operational inference pipelines.

- Developed and evaluated deep learning models (ResNet, VGG) for GNSS interference classification, achieving 98% accuracy across 12 interference types on a dataset of 10,000+ samples. - Built machine learning models for GNSS interference characterization and parameter estimation to support mitigation strategies. - Designed data preparation pipelines, including preprocessing, augmentation, and normalization, improving model robustness and generalization on real-world data.