Traditional CFD-driven design optimization relies on time-consuming manual iterations between simulation and CAD modifications, delaying product development. To overcome this bottleneck, we plan to develop an automated workflow that directly optimizes CAD parameters within a CFD framework, incorporating practical design constraints early in the process. Due to the high-performance requirements of CFD simulations and the sheer number of function evaluations in a multidimensional optimization workflow, establishing this optimization process for product-scale models remains challenging. We plan to overcome these challenges by integrating multiple state-of-the-art techniques into a single workflow capable of tackling even large-scale problems. To ensure an optimal runtime, we address performance challenges on different levels, beginning with the CAD geometry. With the usage of original CAD design parameters instead of mesh-based parameters, we ensure that the optimized shape meets engineering design and manufacturing standards while reducing the dimensionality of the parameter space by orders of magnitude. Furthermore, state-of-the-art techniques such as the discontinuous Galerkin Method implemented in the Trixi.jl [1] framework coupled with the tree-based Adaptive Mesh Refinement capabilities implemented in the t8code [2] library significantly reduce the overall degrees of freedom. Lastly, differentiating the complete workflow, so the CAD geometry, adaptive mesh, and the discontinuous Galerkin solver enables the use of gradient-based optimization methods, which reduce the number of function evaluations in the workflow. Notably, the CAD geometry was already differentiated by Kleinert et al. [3]. The resulting workflow should provide a scalable, automated pathway from CAD to optimized and simulation-verified designs, addressing performance and manufacturability constraints.