Dragos Margineantu

Technical lead of the Artificial Intelligence research. Lead - AI for autonomous systems - perception, planning & decision making, certification. Current focus and interests: - Engineering of AI techniques into pipelines and deployed systems; - Uncertainty estimation and decision making under uncertainty; - Certification of AI systems; design time and runtime assurance methods; - Generative AI methods for decision systems; - Design of systems and approaches that work in the real world; - Developing a strategic vision, hiring and forming highly skilled teams for that vision; - Design, development, testing, validation of computer vision, perception, dialogue, planning for decision systems; - Design of experiments for data collection and testing of high-risk systems; - AI Engineering, building robustness in systems: learning, reasoning, symbolic methods; - Generative models for the design of complex systems (configurations, airplane subsystems, parts); - Sequential decision making - reinforcement learning; - Inverse reinforcement learning. - Anomaly detection and handling surprises at runtime. - Ensemble methods. Deep understanding and extensive experience with machine learning, planning, intelligent decision systems, autonomous ground and air vehicles, airplane health management. Designed and implemented intelligent and learning systems that work in collaborative human-computer environments, surveillance & monitoring and predictive maintenance systems. Sequential decision making methods (reinforcement learning and beyond), anomaly detection, active learning, ensemble methods, cost-sensitive learning, adversarial learning, relational learning. Inverse reinforcement learning & inverse planning + state-action models. Task & motion planning, localization, mapping for autonomous vehicles. Adversarial learning. Intelligent interfacing with domain experts, interactive learn