Automotive aerodynamic design is often slowed by the high cost and long run times of traditional evaluation methods. High-fidelity computational fluid dynamics (CFD) simulations and physical wind tunnel tests can take several days per design iteration, limiting the engineers’ ability to efficiently explore and refine vehicle shapes. This project addresses the challenge by developing a deep learning-based surrogate model for rapid aerodynamic prediction using a multi-fidelity training strategy that combines low-fidelity Reynolds-averaged Navier–Stokes (RANS) data with a smaller set of high-fidelity Improved Delayed Detached Eddy Simulations (IDDES). Once trained, the model will predict key aerodynamic metrics directly from 3D geometry in seconds, without requiring expensive CFD runs. By enabling faster, low-cost aerodynamic analysis, the project will support more energy-efficient vehicle designs, contributing to reduced energy consumption and longer electric vehicle (EV) range. It also exemplifies digital engineering by applying AI to transform traditional vehicle development workflows.