Abdullah Jirjees

Develops computer-vision and image-processing solutions for semiconductor inspection, metrology, and precision imaging with an emphasis on robustness and production usability. Builds and evaluates Python and C++ vision pipelines for preprocessing, detection/segmentation analysis, calibration, validation, and performance characterization. Investigates failure cases tied to imaging variability and data quality, refining algorithms and evaluation procedures for reliable operation. Works with engineering teams to transition R&D methods into maintainable workflows and documentation.

Developed and evaluated deep-learning computer-vision systems for 3D object detection, scene understanding, driver monitoring, and multimodal perception using RGB/IR cameras, LiDAR, and vehicle/physiological signals. Performed dataset preparation, augmentation, model training and fine-tuning, validation, and failure analysis in PyTorch. Used Hugging Face and Detectron2 in project work and integrated models into NVIDIA Jetson Xavier/Orin with ROS/ROS2-based systems for real-time experimentation and deployment validation. Supervised and mentored graduate students and contributed to invention disclosure and technical outputs in multimodal AI.

Developed computer-vision and deep-learning methods for UAV imagery, remote sensing, and geospatial image analysis for field-deployed use cases. Worked with large georeferenced datasets, focusing on registration, coordinate-aware processing, spatial measurements, and quantitative model evaluation. Built Python-based experimental pipelines and collaborated with geomatics teams to translate research into practical image-analysis workflows. Improved robustness by addressing acquisition variability through preprocessing, data curation, model refinement, and validation.

Designed and trained deep-learning vision models using a large, self-collected image corpus exceeding 500,000 images. Built reproducible data preparation, augmentation, training, validation, and benchmarking workflows for large-scale computer-vision experiments. Performed model comparison and error analysis to iteratively improve robustness and generalization on real-world imagery. Managed experimental repeatability and systematic evaluation across runs.

Worked on OCR and assistive vision, AR/VR imaging, real-time nonlinear image recomposition, and medical/military image analysis research. Developed classical and deep-learning image-processing pipelines for tasks such as segmentation-oriented analysis, enhancement, recognition, and real-time vision components. Focused on building end-to-end experimental pipelines for different imaging modalities and performance scenarios. Conducted research experiments to support iterative method refinement.

Researched image segmentation, illumination correction, local contrast enhancement, biomedical imaging, and machine-learning methods, contributing to peer-reviewed publications. Developed MATLAB, Python, and C++ image-processing algorithms with quantitative benchmarking and reproducibility as key evaluation practices. Supported undergraduate teaching in C++ programming and electrical system design, including labs, debugging, and technical guidance. Conducted research-driven experimentation focused on rigorous performance measurement.