Planetary embryos embedded in gaseous protoplanetary disks undergo Type I orbital migration. Migration can be inward or outward depending on the local disk properties but, in general, only planets more massive than several $M_\oplus$ can migrate outward. Here we propose that an embryo's migration history determines whether it becomes a hot super-Earth or the core of a giant planet. Systems of hot super-Earths (or mini-Neptunes) form when embryos migrate inward and pile up at the inner edge of the disk. Giant planet cores form when inward-migrating embryos become massive enough to switch direction and migrate outward. We present simulations of this process using a modified N-body code, starting from a swarm of planetary embryos. Systems of hot super-Earths form in resonant chains with the innermost planet at or interior to the disk inner edge. Resonant chains are disrupted by late dynamical instabilities triggered by the dispersal of the gaseous disk. Giant planet cores migrate outward toward zero-torque zones, which move inward and eventually disappear as the disk disperses. Giant planet cores migrate inward with these zones and are stranded at ~1-5 AU. Our model reproduces several properties of the observed extra-solar planet populations. The frequency of giant planet cores increases strongly when the mass in solids is increased, consistent with the observed giant exoplanet - stellar metallicity correlation. The frequency of hot super-Earths is not a function of stellar metallicity, also in agreement with observations. Our simulations can reproduce the broad characteristics of the observed super-Earth population.