This talk presents a unified vision for designing and securing AI-enabled software systems through the integration of two complementary research threads: a semantic-driven requirements engineering framework for AI perception systems and a predictive framework for modeling cyberattack-defense co-evolution in software. The first part introduces a semantic-aware requirements engineering approach tailored for AI-based perception systems that inherently lack explicit specifications. By mining structured domain knowledge from textual and visual sources, the framework constructs ontologies that define semantic expectations for critical concepts (e.g., "pedestrian," "aircraft"). These specifications are then used to evaluate and iteratively retrain AI models and their training datasets, measuring perceptual alignment through explainable AI techniques. This process enhances concept recognition, improves specification fidelity, and promotes generalizable performance, ensuring that the models' behavior more closely reflects real-world expectations. The second part of the talk focuses on a data-driven framework that mines software repositories to uncover and leverage attack-defense co-evolution patterns. By analyzing historical vulnerability-patch-attack triplets through multimodal embeddings of source code and natural language artifacts, the framework constructs predictive models that anticipate potential attacker adaptations to defensive measures. This capability supports proactive security planning and lays the groundwork for developing self-robusting software—systems that continuously evolve to preemptively defend against emerging threats.

Bio: Dr. Mona Rahimi is an Associate Professor of Software Engineering at Northern Illinois University. She earned her Ph.D. from the University of Notre Dame, focusing on software engineering for safety-critical systems. Her current work emphasizes AI-driven software engineering, AI engineering, and AI-based cybersecurity. At NIU, she serves as the director of the RAiSE Lab (Reliable AI-based Software Engineering) and is currently leading projects funded by the National Science Foundation, the Department of Defense, and multiple industry partners. She maintains active collaborations across both government and industrial sectors.