This paper presents a new approach to the resource-constrained project scheduling problem (RCPSP) under uncertainty, using a scenario-based method aimed at minimizing worst-case performance through a min-max robustness objective. The proposed GRASP (Greedy Randomized Adaptive Search Procedures) method incorporates an adaptive greedy function and local search heuristic for effective solution construction and improvement, demonstrating superiority over basic procedures in robustness optimization on benchmark datasets. Experimental results indicate that the GRASP approach effectively manages the complex nature of RCPSP while addressing activity duration uncertainty.