This work proposes a new and simple algorithm for unconstrained numeric optimization over continuous spaces. A population of candidate solutions styled as a herd of kudus performs a succession of jumps through the search space in order to find the best solution (the kudu is a species of antelope). The logic of this algorithm is quite different from that of most populationbased algorithms, as the individual solutions are moved together in a herd-like fashion. Performance comparisons are conducted with the Artificial Bee Colony, Differential Evolution, the Genetic Algorithm and Particle Swarm Optimization on benchmark functions. The kudu herd seems to perform well in the early stages and on high-dimensional problems.