<p>This work presents a gradient-based optimization framework for parameterized CAD models incorporating both geometric and composite material design variables. Traditional sequential optimization approaches for shape and material properties often yield sub-optimal designs due to their inability to capture coupled effects, while topology optimization methods lack direct integration with editable CAD representations. To address these limitations, the proposed method combines efficient adjoint sensitivity analysis for physical derivatives and general finite differences for geometric CAD derivatives, enabling simultaneous optimization of heterogeneous CAD parameters (e.g., control points, fiber orientations) and composite material distributions. The present workflow maps CAD geometries to intermediate material fields, computes gradients using the chain rule, and updates the design parameters using efficient gradient-based algorithms. Case studies on 2D and 3D structures demonstrate that concurrent optimization of shape and anisotropic material properties improves the mass and stiffness of the editable CAD designs. The optimization results highlight how the fiber continuity and surface-conformality constraints influence the optimized designs while preserving or enhancing their manufacturability. Computational performance analysis reveals that the finite element simulation remains the primary runtime cost, similarly to efficient topology optimization workflows. Hence, the present work bridges the gap between industrial CAD workflows and structural composite optimization, enabling efficient exploration of coupled design spaces.</p>