What particle shape will generate the highest packing fraction when randomly poured into a container? In order to explore and navigate the enormous search space efficiently, we pair molecular dynamics simulations with artificial evolution. Arbitrary particle shape is represented by a set of overlapping spheres of varying diameter, enabling us to approximate smooth surfaces with a resolution proportional to the number of spheres included. We discover a family of planar triangular particles, whose packing fraction of φ ∼ 0.73 outpaces almost all reported experimental results for random packings of frictionless particles. We investigate how φ depends on the arrangement of spheres comprising an individual particle and on the smoothness of the surface. We validate the simulations with experiments using 3D-printed copies of the simplest member of the family, a planar particle consisting of three overlapping spheres with identical radius. Direct experimental comparison with 3D-printed aspherical ellipsoids demonstrates that the triangular particles pack exceedingly well not only in the limit of large system size but also when confined to small containers.The relationship between particle shape and packing density for random particle arrangements continues to be a topic of considerable interest. 1-3 Going beyond spherical particles, there has been much recent progress for polyhedral or polygonal shapes, 4-9 ellipsoids, cuboids, or 'superballs,' 10-15 cylinders, cones, and frustums of different aspect ratios, [16][17][18] as well as various types of particles constructed by joining disks or spheres. 11,[19][20][21][22][23][24][25] Furthermore, in the last few years, increasing attention has been paid to particles that are highly non-convex or are sufficiently flexible to assume non-convex shapes during the packing process. 14,[26][27][28][29][30][31][32][33][34] In almost all cases, these studies proceeded from a given particle type to find the packing density. What has remained a major challenge is a general and systematic approach to the inverse problem: taking desired packing properties as a starting point to identify the appropriate particle shape.To make progress, it is necessary to address three main obstacles. Firstly, shape is an infinitely variable parameter, rendering an exhaustive investigation of all possible particle shapes infeasible and instead requiring a smart approach to quickly narrow the search. The second obstacle relates to the fundamental nature of amorphous, jammed aggregates. Because a jammed system exists far from equilibrium, the packing density φ can be affected not only by particle geometry but also by boundary and processing conditions. These effects become particularly pronounced when extrapolating from finite size experiments and simulations to the properties of an infinite system. Finally, different shapes may generate similar packing fractions when assembled into an aggregate under particular processing conditions, so that for a given level of approximation to the desired target density ...