Community detection is a fundamental task in complex network analysis, enabling the identification of hidden structures and functional groupings within social, biological, and technological systems. Although multi-objective genetic algorithms have been shown to perform well for this task, their practical application is typically constrained by excessive computational expense, especially on large or dense networks. In this paper, we present an efficient hybrid genetic algorithm for multi-objective community detection that makes a good trade-off between solution quality and computational efficiency. Our algorithm incorporates several essential improvements, including a fitness caching technique to prevent duplicate evaluations, a lightweight crossover operator to minimize the overhead of the evolutionary process, and an efficient archive management strategy that prioritizes high-quality solutions along the Pareto front. Experimental evaluations on several benchmark datasets demonstrate that the proposed method achieves comparable or improved modularity (Q) values, maintains stable normalized mutual information (NMI) values, and significantly reduces execution time compared with the previously proposed method, achieving up to 57.7% runtime reduction on large-scale networks. The results confirm that the proposed method provides a viable and scalable approach to community detection in large-scale complex networks.