Breakthroughs in heat sink design are now limited by conventional manufacturing techniques which can only produce basic fin shapes with homogenous surface structuring. Additive Manufacturing (AM) is a potential disruptive technology for heat sink design and other convective heat transfer solutions, as surface properties can be directionally varied to maximize performance for a prescribed environment and flow structure. This study examines the thermal and aerodynamic performance effects of surface roughness variation along the height and length of plane fin heat sinks. Pyramid structures have been selectively applied to the surface of the sink to produce uniform surface roughnesses of 5, 15, 30, 43 and 71 microns on heat sinks that have been created to have a roughness gradient in directions parallel and perpendicular to ambient flow. In anticipation of applicability to high air speed aerospace environments, the experimental facility is placed in the center of a wind tunnel in order to expose the sink to the maximum possible velocity, which causes the flow structure to be considered an open flow. A plane flat surface and a heat sink with a uniform surface roughness of 5 microns, validated by a white light interferometer, are compared to theoretical expectations found using Nusselt number correlations for turbulent flow over a flat plate for the flat surface, as well as the Gnielinski correlation and the Colburn equation for the heat sinks. A thermal conductivity of 11.9 W/m-K is found using a Thermal Interface Material (TIM) tester, showing a significant reduction in this property relative to the expected property of bulk aluminum which has been conventionally machined. This reduction is due primarily to the porosity of the internal structuring which is an artifact of the Selective Laser Melting (SLM) process used to fabricate the heat sinks. Using a 7-plane fin heat sink where each 2.5 mm (0.1 in.) thick fin is spaced 3.3 mm (0.13 in.) apart, a minimum thermal resistance of 0.23 K/W is found. Graded roughness in the direction of the flow is found to have no effect on thermal resistance, however, gradients in the direction normal to the bulk flow are found to decrease the thermal resistance by 3.5% for a Reynolds number range of 8,300 to 28,700 where the Reynolds number is based on the characteristic length of double the spacing between the fins. Improvements in thermal resistance reach as high as 15.2% when comparing heat sink with full maximum surface roughness patterns distributed evenly across the fins, coming at a weight increase of 4-12%. Intelligent heat sink design which exploits the surrounding flow conditions can improve performance drastically with minimal increase to heat sink weight and volume. However, this functionally graded surface roughness parameter is only possible through AM techniques. Positive impacts on performance from surface roughness gradients introduces the possibility of finding AM surface roughness patterns which are tuned using numerical modelling of the surrounding flow structure and hydrodynamic boundary layer development.