Balancing influence maximization with demographic fairness remains a critical challenge in recommendation systems. Existing methods either ignore topic-sensitive fairness or rely on heuristic approximations that lack scalability. We propose Fair2Vec, a novel framework to jointly recommend (1) top-k influencers, (2) top-r topics, and (3) ensure the influenced population’s demographic distribution aligns with the broader topic-specific community. Fair2Vec leverages topic-aware embeddings to model influence dynamics and fairness constraints, eliminating error-prone multi-hop computations. By restructuring networks as bipartite graphs, it reduces time complexity compared to state-of-the-art heuristics. Experiments on Meetup, Yelp, and DBLP datasets demonstrate Fair2Vec’s superiority: it achieves higher fairness and greater influenced populations than baselines. Our work bridges the gap between scalable influence maximization and topic-aware fairness in recommendations, offering actionable insights for platforms aiming to foster inclusive engagement while preserving relevance to user interests.