Toward a practical application of the additive manufacturing (AM), this study proposes a shape optimization approach for the cross-sectional shape of the inner pipe of a counter-flow double pipe heat exchanger. The cross-sectional shape of the inner pipe is expressed by an algebraic expression with a small number of parameters, and their heat transfer performance is evaluated by a commercial Computational Fluid Dynamics (CFD) solver. The optimization is conducted by the Non-Dominated Sorting Genetic Algorithm II (NSGA-II) assisted by the Kriging surrogate model, and the NSGA-II finds the optimal cross-sectional shape with many protrusions around the perimeter of the inner channel to improve the heat transfer performance. In this study, heat transfer performance is evaluated from the temperature drop at the outlet of the high-temperature fluid. Through the comparison of two cross-sectional shapes with the same heat transfer surface area — average temperature at the outlet of the optimal high-temperature channel is 324.58 K while average temperature at the outlet of a circular high-temperature channel with the same area as the optimal channel is 331.93 K, it is revealed that the number of protrusions plays important roles which contribute not only to increase heat transfer area but also to improve heat transfer performance.