Enzymes collectively display a great breadth of catalytic properties yet are individually confined to one or a few specific catalytic tasks. Despite key advances in enzyme engineering, our capacity to predict the effects of mutations on function remains nebulous. Here we present advances in engineering non-native biocatalyzed transformations to procure useful products. We focus on cytochrome P450 oxidase from Bacillus megaterium (P450 BM3) in its capacity to functionalize unactivated C-H bonds. We present cost-effective, high-throughput colorimetric screening at the whole-cell level to streamline discovery of new reactivity in panels of hundreds of P450 BM3 variants. We greatly expand the diversity of P450 BM3 variants that display promiscuous aromatic hydroxylation and epoxidation reactivity. We look ahead to the potential for large experimental datasets to train design algorithms for enzyme engineering.