The natural microbial functions of many soils are severely degraded. Current state-of-the-art technology to restore these functions is through the isolation, screening, formulation and application of microbial inoculants and synthetic consortia. These approaches have inconsistent success, in part due to the incompatibility between the biofertilizer, crop, climate, existing soil microbiome and physicochemical characteristics of the soils. Here, we review the current state of the art in biofertilization and identify two key deficiencies in current strategies: the difficulty in designing complex multispecies biofertilizers and the bottleneck in scaling the production of complex multispecies biofertilizers. To address the challenge of producing scalable, multispecies biofertilizers, we propose to merge ecological theory with bioprocess engineering to produce 'self-assembled communities' enriched for particular functional guilds and adapted to a target soil and host plant. Using the nitrogen problem as an anchor, we review relevant ecology (microbial, plant and environmental), as well as reactor design strategies and operational parameters for the production of functionally enriched self-assembled communities. The use of self-assembled communities for biofertilization addresses two major hurdles in microbiome engineering: the importance of enriching microbes indigenous to (and targeted for) a specific environment and the recognized potential benefits of microbial consortia over isolates (e.g. functional redundancy). The proposed community enrichment model could also be instrumental for other microbial functions such as phosphorus solubilization, plant growth promotion or disease suppression.