Uniform wafer temperature control is essential in semiconductor manufacturing, where small thermal variations can degrade process quality. This study presents a computational optimization framework for the geometric design of susceptor heating elements to improve temperature uniformity. A circular zig-zag heating coil geometry is considered, offering high geometric flexibility for large-area susceptors with spatially varying heat losses. Each coil is parameterized by pattern density, radial position, and height. Steady-state heat transfer is evaluated using a fully automated three-dimensional finite element analysis framework. To efficiently address the resulting high-dimensional and mixed-variable design space, Gaussian process–based Bayesian optimization is employed. A stage-wise optimization strategy is introduced, in which subsets of design variables are optimized sequentially to enhance convergence efficiency. The method is applied to susceptor designs with multiple concentric heating coils. Numerical results demonstrate that the stage-wise approach significantly improves temperature uniformity while reducing the number of expensive finite element evaluations. The proposed framework provides an efficient computational strategy for complex thermal design problems in semiconductor processing. Acknowledgement This work was supported by the Technology Innovation Program (20026346, Development of susceptor for electrostatic chuck with reduced particle generation more than 50 % under plasma condition) funded by the Ministry of Trade Industry & Energy (MOTIE, Korea).