Text-Based Person Search (TBPS) aims to retrieve person images from a gallery using natural language descriptions. This task remains challenging due to the heterogeneous nature of visual and textual data, along with inter-identity ambiguity and intra-identity variations. To address these issues, we present MAMBA-TBPS, an efficient and attribute-aware framework that integrates five complementary objectives: contrastive loss for modality alignment, relation-aware loss for semantic matching, sensitivity-aware loss for capturing fine-grained word-level cues, attribute loss for emphasizing identity-specific traits, and a masked visual reconstruction loss to enhance visual feature learning. This unified training strategy allows the model to better align and interpret cross-modal data. Extensive experiments on three benchmark datasets—CUHK-PEDES, ICFG-PEDES, and RSTPReid—demonstrate that our method achieves competitive performance. Code is publicly available at: https://github.com/Mehulk43/MAMBA-TBPS .