Winglets on a turbine blade can modify flow features and improve a marine current turbine (MCT). In this work, to maximize the power coefficient (CP) and torque (T), cant angle (α), and height (h) of a winglet of an MCT were modified. The problem was solved using a high-fidelity solver code Fluent 2019R2 containing Reynolds-Averaged Navier–Stokes (RANS) equations. The flow domain meshed with tetrahedral elements. Nine different designs were produced to fill the design space for optimization. A set of low fidelity models such as second-order regression, kriging, and neural network models were used to approximate the high-fidelity results. The optimal designs further validated with the high-fidelity simulated results. The optimal design, increased CP by 7% for α = 33.2o and h = 2.04% of the turbine radius, reduced the recirculation zone at the trailing edge, increased the pressure gradient, and reduced the tip vortex.