TRAIN lab is dedicated to advancing the frontiers of trustworthy, resilient, and generalizable AI methods for networked and cyber-physical systems, with an emphasis on next-generation wireless networks such as 6G. We explore foundational questions at the intersection of machine learning, wireless communication, and control theory, aiming to enable AI systems that are not only high-performing but also interpretable, robust to uncertainty, and aligned with real-world constraints. Key areas include explainable AI for network decision-making, AI-native wireless architectures, neuroscience-inspired world modeling, and resilience in distributed intelligent systems. Our approach combines theoretical insights with practical validation to help shape the future of safe and deployable AI in complex environments.