As a primary interface between adaptive and innate immune responses, T cells serve as a central node in regulating the immune system. In turn, T cell response is governed by antigen recognition via the T cell receptor (TCR). This recognition and subsequent response coordinate a series of transcriptional programs. Single-cell RNA and paired TCR profiling offer insights into numerous physiological and pathological processes. Unlike the plethora of single-cell RNA analysis pipelines, computational tools that leverage single-cell TCR sequences for further analyses are wanting. We developed a deep learning-based approach to transform complementarity-determining region 3 (cdr3) amino acid sequences into vectors using the latent dimensional space. RNA expression and TCR sequences can be co-embedded with these transformed values to produce a dimensional reduction that encompasses an immune response. We then applied this approach to look at spatiotemporal T follicular helper response in COVID-19 vaccinations and identify candidate TCRs that bind the spike antigen of SARS-CoV2. Our confirmatory studies cloning chimeric TCRs into Jurkat cells are the first in vitro confirmed deep learning-based TCR antigen predictions.