Highâthroughput tissue barrier models can yield critical insights on how barrier function responds to therapeutics, pathogens, and toxins. However, such models often emphasize multiplexing capability at the expense of physiologic relevance. Particularly, the distal lung's airâblood barrier is typically modeled with epithelial cell monoculture, neglecting the substantial contribution of endothelial cell feedback in the coordination of barrier function. An obstacle to establishing highâthroughput coculture models relevant to the epithelium/endothelium interface is the requirement for underside cell seeding, which is difficult to miniaturize and automate. Therefore, this paper describes a scalable, lowâcost seeding method that eliminates inversion by optimizing medium density to float cells so they attach under the membrane. This method generates a 96âwell model of the distal lung epitheliumâendothelium barrier with serumâfree, glucocorticoidâfree airâliquid differentiation. The polarized epithelialâendothelial coculture exhibits mature barrier function, appropriate intercellular junction staining, and epithelialâtoâendothelial transmission of inflammatory stimuli such as polyinosine:polycytidylic acid (poly(I:C)). Further, exposure to influenza A virus PR8 and human betaâcoronavirus OC43 initiates a doseâdependent inflammatory response that propagates from the epithelium to endothelium. While this model focuses on the airâblood barrier, the underside seeding method is generalizable to various coculture tissue models for scalable, physiologic screening.