Biological signaling is often governed by extreme value statistics, where a rapid response relies on the fastest few out of a large redundant group of searchers. While extreme first passage time (FPT) theory is well established for homogeneous ensembles, its sensitivity to population heterogeneity remains open. We show that averaging over a heterogeneous population of memoryless random walkers gives rise to ensemble self-reinforcement. We obtain closed-form expressions for the ensemble-averaged occupation probability, FPT density, and extreme arrival statistics of a heterogeneous population of biased continuous-time random walkers with gamma-distributed rates and beta-distributed biases. The heterogeneity drastically changes both the FPT and minimum FPT densities relative to a homogeneous ensemble with identical mean rates. For heterogeneous populations, the modes of the FPT and minimum FPTs are an order of magnitude smaller than for homogeneous ones. Our exact analytical predictions establish that population heterogeneity is a parameter that biology can exploit and not merely noise to be averaged away.