The development of high-power electronic devices in various energy systems has recently gained attention for electric vehicles and hybrid electric vehicles. However, the increase in power density has also introduced several design challenges especially in thermal management. This thesis attempts to optimize a liquid cooled pin fin heat sink with localized heat sources representing an EV power driver using genetic algorithm (GA). Both pin geometry and pin locations are optimized to reduce thermal resistance and pressure drop of the coolant. Computational fluid dynamics (CFD) simulation is used in the genetic algorithm to evaluate the performance of each potential design during optimization. A set of optimizations with different geometric parameters has been carried out and compared. The results demonstrate that GA coupled with CFD may be a powerful tool for designing an effective and reliable heat sink for high power density applications.