Some active fault systems comprise near-orthogonal conjugate strike-slip faults, as highlighted by the 2019 Ridgecrest and the 2012 Indian Ocean earthquake sequences. In conventional failure theory, orthogonal faulting requires a pressure-insensitive rock strength, which is unlikely in the brittle lithosphere. Here, we conduct 3D numerical simulations to test the hypothesis that near-orthogonal faults can form by inheriting the geometry of deep ductile shear bands. Shear bands nucleated in the deep ductile layer, a pressure-insensitive material, form at 45^o from the maximum principal stress. As they grow upwards into the brittle layer, they progressively rotate towards the preferred brittle faulting angle, ~30^o, forming helical shaped faults. If the brittle layer is sufficiently thin, the rotation is incomplete and the near-orthogonal geometry is preserved at the surface. The preservation is further facilitated by a lower confining pressure in the shallow portion of the brittle layer. For this inheritance to be effective, a thick ductile fault root beneath the brittle layer is necessary. The model offers a possible explanation for orthogonal faulting in Ridgecrest, Salton Trough, and Wharton basin. Conversely, faults nucleated within the brittle layer form at the optimal angle for brittle faulting and can cut deep into the ductile layer before rotating to ~45^o. Our results thus reveal the significant interactions between the structure of faults in the brittle upper lithosphere and their deep ductile roots.