The hyper-production of β-glucosidase by a local strain of Beauveria bassiana under submerged conditions is reported in this study. The initial screening of seven agricultural residues showed that the haulm of Bambara—an underutilized African legume—supported the highest β-glucosidase production; hence, statistical optimization of enzyme production was done using this biomass as the sole carbon source. Plackett–Burman design identified the concentrations of Bambara haulm, KCl, and NaCl as well as agitation speed and incubation time as the most significant factors affecting enzyme production. Subsequently, the central composite design predicted the optimal conditions (Bambara 57 g/L, KCl 302 mg/L, NaCl 154 mg/L, agitation speed 150 rpm, and incubation 223 h) for B. bassiana β-glucosidase production, which were further validated. The generated quadratic model was deemed significant judging from its F-value (201.63), adequate precision ratio (45.74), as well as the R2 (0.9988), adjusted R2 (0.9938), and predicted R2 (0.9195) values. The optimization resulted in a ~5.36-fold increase in enzyme levels from the unoptimized production of ~133 to 711 U/mL. The enzyme was also demonstrated to efficiently hydrolyze cellobiose, converting more than 90% of the substrate to glucose. These results further establish the resourcefulness of the B. bassiana strain for the production of β-glucosidase enzyme, having immense potential, especially in the food and energy industries.