A mechanistic understanding of host-microbe interactions in the gut microbiome is hindered by poorly annotated bacterial genomes. While functional genomics can generate large gene-to-phenotype datasets to accelerate gene discovery, their applications to study gut anaerobes have been limited. For instance, most gain-of-function screens of gut bacterial genes have been performed in an aerobic host and included a small number of conditions. To address these challenges, we developed a strategy to barcode expression libraries for high-throughput interrogation of gene functions in competitive fitness assays. We demonstrate the power of this approach to uncover novel phenotypes for uncharacterized genes using pooled libraries constructed from a diverse set of gut Bacteroidales expressed in Bacteroides thetaiotaomicron. We identified new roles in carbohydrate metabolism for nine proteins, including enzymes, transporters, a regulator, and hypothetical proteins from mobile genetic elements. This approach can be readily applied to other organisms and additional phenotypic assays.