We confirm and characterize a close-in (P-orb = 5.425 days), super-Neptune sized (5.04(-0.37)(+0.34) R-circle plus) planet transiting K2-33 (2MASS J16101473-1919095), a late-type (M3) pre-main-sequence (11 Myr old) star in the Upper Scorpius subgroup of the Scorpius-Centaurus OB association. The host star has the kinematics of a member of the Upper Scorpius OB association, and its spectrum contains lithium absorption, an unambiguous sign of youth (<20 Myr) in late-type dwarfs. We combine photometry from K2 and the ground-based MEarth project to refine the planet's properties and constrain the host star's density. We determine K2-33's bolometric flux and effective temperature from moderate-resolution spectra. By utilizing isochrones that include the effects of magnetic fields, we derive a precise radius (6%-7%) and mass (16%) for the host star, and a stellar age consistent with the established value for Upper Scorpius. Follow-up high-resolution imaging and Doppler spectroscopy confirm that the transiting object is not a stellar companion or a background eclipsing binary blended with the target. The shape of the transit, the constancy of the transit depth and periodicity over 1.5 yr, and the independence with wavelength rule out stellar variability or a dust cloud or debris disk partially occulting the star as the source of the signal; we conclude that it must instead be planetary in origin. The existence of K2-33b suggests that close-in planets can form in situ or migrate within similar to 10 Myr, e.g., via interactions with a disk, and that long-timescale dynamical migration such as by Lidov-Kozai or planet-planet scattering is not responsible for all short-period planets.