This study develops a model aimed at optimizing workforce schedules in a dark kitchen company, an environment characterized by high demand variability, peaks concentrated during lunch and dinner periods, and the coexistence of different labor regimes. To translate the stochastic behavior of orders into concrete operational requirements, the upper tolerance limit of the estimated working time (UTL of EOT) was employed, based on historical order data. This indicator made it possible to estimate, with statistical confidence, the minimum staffing requirement for each hour and day of the week. Based on these parameters, an integer linear programming model was formulated, whose objective function seeks to minimize the total monthly labor cost, while satisfying operational and legal constraints such as minimum coverage per time interval, mandatory presence of permanent employees, and admissible shift start windows for each employment regime.