Inverse modeling was applied to the optimization of a crucible design for SiC sublimation growth. We found a crucible shape providing the optimal temperature distribution in terms of the powder source stability during long-term operation and of the convex crystal shape. Considerable improvement of temperature uniformity throughout the powder charge was achieved. The results obtained show selective sensitivity of the thermal field inside the crucible to modification of the crucible design. The inverse problem approach is easy-to-adapt to various optimization criteria and seems to be especially effective in the case of multi-factor optimization.