The accelerating demand for design of sustainable, efficient, and high-performance vehicles is driving a transformation in engineering methodologies. Among the most promising enablers of this transformation is machine learning (ML), which offers powerful tools for reducing development time, accelerating design space exploration, and enabling real-time decision-making across multiple fidelity levels (Serani et al. 2024, Di Fiore et al. 2024). This Minisymposium aims to gather contributions that explore the integration of ML into vehicle design workflows, with particular focus on early-stage and multidisciplinary processes. Targeted applications include air, sea, and ground vehicles, spanning a broad range of mobility systems. Topics of interest include, but are not limited to: (i) ML-based surrogate modeling and dimensionality reduction for complex, high-dimensional design spaces; (ii) Reduced-order modeling integrated with ML to replace or augment high-fidelity solvers; (iii) Adaptive sampling and optimization strategies across fidelity levels; (iv) Multi-fidelity modeling and data fusion approaches; (v) Explainable and physics-informed ML methods to ensure design transparency and accountability; (vi) ML-driven design methodologies with a focus on sustainability metrics; (vii) Benchmarking and best practices for ML-enhanced design frameworks. The Minisymposium is motivated by active international collaborations investigating how ML can improve performance, robustness, and sustainability of vehicle design processes. By bridging communities working on ML, reduced-order modeling, optimization, and computational mechanics, the session seeks to foster critical dialogue, highlight emerging trends, and build synergies across disciplines and application domains (Mendez et al., 2025).