Permafrost thaw, coastal erosion, flooding, and ground instability are creating escalating risks for infrastructure and communities in Arctic and other cold-region environments. This session invites contributions that apply artificial intelligence and machine learning to characterize, monitor, and predict infrastructure vulnerability, hazard evolution, and system performance under rapid environmental change. We welcome studies using physics-informed machine learning, hybrid process-data models, Bayesian inference, surrogate modeling, uncertainty quantification, anomaly detection, pattern recognition, data fusion, emerging digital twin approaches, and decision-support tools. Relevant applications include roads, airstrips, pipelines, buildings, coastal protection systems, and community infrastructure exposed to interacting thermal, hydrologic, geomorphic, and geotechnical processes. Contributions may also address integration of field observations, geophysical and geotechnical monitoring, and mechanistic modeling. We especially encourage interdisciplinary studies that connect process understanding with actionable tools for resilience planning, adaptation, infrastructure management, and community decision-making in permafrost and Arctic coastal settings.