Given the rapidly evolving and competitive labor market, E-recruitment recommendation systems have gained increasing attention for their ability to leverage comprehensive user profile data and deliver personalized job recommendations. However, research addressing recommendation algorithms specifically for career-break populations, who require the most targeted career guidance, remains unexplored. Therefore, this quantitative study aims to investigate the technical challenges underlying this research gap to improve recommendation accuracy: how to incorporate career break information and how to handle varying profile sequence lengths among career break users. To address these challenges, we propose a hybrid Graph Neural Network (GNN) approach for sequence-aware job recommendation. Our two-stage framework combines a GNN-based ranking stage with a dynamic gating rule–based re-ranking stage that adaptively balances industry and skill signals through a sigmoid-based gating mechanism. Our experiments are conducted on a dataset, containing 2,238 users with 12 distinct career break categories. Experimental evaluation demonstrates that our proposed system achieves Hit Rates of 37.2% at Top 10, 49.6% at Top 30, and 71.6% at Top 50, highlighting its effectiveness in addressing the career break job recommendation challenge.