An optimization method is proposed to design a dc-dc boost converter, including the differential input filter and the associated control strategy, for embedded applications [aerospace, electric vehicle/hybrid electric vehicle (EV/HEV)], where high power density and high efficiency are strongly required. The problem is treated as a multiobjective optimization, whose purpose is to maximize compactness and efficiency of the converter. In order to find a tradeoff between the two objective functions (volume and power losses), a genetic algorithm was employed (NSGA-II) to generate the most convenient design solutions. The operating conditions includes both steady-state and load power transients of a constant power load (CPL). In order to comply with the specification on the output voltage admitted range in both conditions, the control strategy is discussed and included in the design procedure. The interaction of a differential input filter with a CPL may lead to serious instability issues. A constraint function in the algorithm excludes all the unstable and unfeasible design solutions. A high-efficiency converter ( $\eta =98.5\%$ ) was realized, and experimental tests prove that stability and design specifications are respected.