This paper addresses a resilient humanitarian service network design against hurricane disruptions for a major carrier which operates a feeder hub-and-spoke network in the Caribbean. The significance of a resilient transportation network- either a public or a private network- arises when a transportation network is utilized for humanitarian purposes immediately after the disaster. The carrier aims to combine non-urgent relief supplies with regular business cargo to achieve transportation efficiency as well as meeting the humanitarian need of the affected areas in the post-disaster period. We propose a two-phase solution framework which is composed of two distinct models, the outer model and the inner model in each corresponding phase. We develop a multi-stage stochastic programming formulation for solving the resilient humanitarian network design problem without scheduling decisions in the outer model. The inner model considers the service network design problem on a time-space network where routing and scheduling decisions are made simultaneously. The output of the outer model-amount of business and relief cargo assigned to each route- is employed as an input for the inner model.