The hole type is the key way to improve the film cooling effectiveness. Nekomimi film hole has good cooling performance but complicated configuration. The Isight software was used to optimize Nekomimi film hole through parametric modeling. Variable sensitivity was conducted to reduce variable dimensions and potential value ranges. The Latin hypercubic and the uniform design methods were both used to construct design sample space to improve the prediction accuracy of the surrogate model. Results show that the deviation of cooling efficiency between the surrogate model and the simulated computation is within 1.0%. An optimal hole obtained via the multi-island genetic algorithm has significant 85.7% improvement of surface averaged film cooling efficiency over the reference hole. The surface averaged cooling efficiency of the optima hole is increased by 76% compared to the reference hole, and the laterally and the centerline cooling efficiency are consistant with numerical results.