T cells are instrumental in orchestrating cell-mediated adaptive immunity by their unique ability to recognize antigens via T cell receptors (TCRs) and initiate downstream immune responses. TCRs exhibit an astonishing degree of variability, surpassing numbers of T cells within an individual. However, only a minute fraction of T cells possesses the capacity to recognize a specific antigen, playing a pivotal role in particular diseases. The precise identification of T cell subsets with specificity to an antigen within the vast T cell repertoire holds immense promise for advancing research in fields ranging from infectious diseases and autoimmunity to vaccine development and cancer immunotherapy. This study proposes an innovative approach that integrates TCR sequences with transcriptional profiles to discern T cells that are not only transcriptionally active during immune responses but also possess convergent antigen-recognition capabilities. Leveraging the capabilities of cutting-edge biology foundation models, our methodology involves encoding the transcriptional profiles and TCR sequences of T cells. Subsequently, we will employ a contrastive learning strategy to align these two modalities within a novel latent space, thereby creating a more representative depiction of cell nature. Within this new latent space, we will cluster the T cells to construct an antigen-reactivity network. Through this research, we aim to gain deeper insights into the T cell response to various diseases. Furthermore, we seek to identify potential TCR candidates suitable for personalized adoptive T cell therapies. This interdisciplinary approach promises to revolutionize our understanding of T cell behavior, offering new avenues for precision medicine and therapeutic development.