Ocean surface boundary layer turbulence plays pivotal roles in shaping the oceanic environment and influencing Earth's climate dynamics. Despite their significance, these fine-scale ocean currents can not be simulated ocean and climate models and are approximated by simplified formulas call parameterizations. Traditionally, parameterizations are derived solely from fundamental physics principles. In this talk, I will present our recent efforts using machine learning techniques to improve those parameterizations and to apply the machine learning based parameterization to better under ocean surface boundary layer turbulence.