Waterlogging significantly reduces crop yield and quality, but its molecular effects on cowpea ( Vigna unguiculata (L.) Walp.) across different developmental stages remain poorly understood. Employing RNA-seq, this study investigates the transcriptomic responses of two contrasting cowpea genotypes, UCR369 and Epicselect.4, under control and waterlogging conditions during the vegetative, flowering, and maturity stages. Differential expression analysis revealed stage-specific and genotype-dependent transcriptional responses to waterlogging, with UCR369 demonstrating more dynamic adjustment, particularly at flowering and maturity, highlighting its improved regulatory plasticity under stress. KEGG pathway enrichment analysis showed increases in phenylpropanoid, flavonoid, and isoflavonoid metabolism, starch-sucrose metabolism, and cutin-suberin-wax biosynthesis, which were most prominent in UCR369 during flowering and maturation, indicating a genotype-specific change in metabolic defense and cell-surface/barrier remodeling. Pathview maps also revealed continuous activation of core flavonoid and isoflavonoid steps throughout development stages in both genotypes, implying an antioxidant and membrane-protective program during waterlogging. Transcription factor activity was tracked longitudinally across development stages: MYBs showed larger expression changes at the vegetative stage, bHLHs at flowering and maturation, and ERFs throughout; this matches up with phenylpropanoid, carbon partitioning, and hormone/redox control, along with additional ROS- and hormone-integrating families (WRKY, HD-ZIP, HSF) identified in our TF clustering. WGCNA detected 35 modules across the genes and developmental stages: the greenyellow module was associated with starch and phenolics production and was enriched for plastid/redox hub genes; coral3 monitored sucrose metabolism and contained genes for sugar interconversion and senescence; firebrick3/bisque4 were associated with plastid and pigment functions that matched photosynthetic adjustment; and brown4 was related to membrane stability (MSI). XGBoost machine-learning analysis condensed these transcriptome-scale patterns into a compact collection of predictor genes that recapitulated the KEGG and WGCNA signals and covered major functional axes including cuticle and membrane integrity, plastid and redox homeostasis, ethylene and senescence signaling, and yield determination; this panel consisted of LTP3, NIA2/NR2, CER1/CER22, CASPL1D1, HSP21, ACS6, ICL, SAG12, ATDI21/DI21, ACO1, and two uncharacterized loci, Vigun07g271600 and Vigun08g155100. Together, these findings reveal the architecture of a multilayered regulatory scheme for waterlogging tolerance, as well as suggesting practical gene targets/markers for breeding more resilient cultivars.