In this article, a design optimization technique for mixing in a gas turbine combustor is presented. The technique entails the use of computational fluid dynamics and mathematical optimization to optimize the combustor exit temperature profile. Combustor geometric parameters were used as optimization design variables. This work does not intend to suggest that combustor exit temperature profile is the only performance parameter important for the design of gas turbine combustors. However, it is a key parameter of an optimized combustor that is related to the power output and durability of the turbine. The combustor in this study is an experimental liquid-fuelled atmospheric combustor with a turbulent diffusion flame. The computational fluid dynamics simulations use a standard k-epsilon model. The optimization is carried out with the Dynamic-Q algorithm, which is specifically designed to handle constrained problems where the objective and constraint functions are expensive to evaluate. The optimization leads to a more uniform combustor exit temperature profile than with the original one.