In this study, we identified and screened nine novel umami peptides derived from yellowfin tuna (Thunnus albacares) utilizing peptidomics combined with various machine learning based umami screening methodologies, including UMPred-FRL, TPDM, Umami-MRNN, and iUmami-SCM. The results of the multisensory evaluation indicated that SI-5 and VE-8 exhibited a high intensity of umami flavor, while AP-6, LD-5, and LT-5 effectively masked bitterness without compromising umami intensity. Furthermore, molecular docking techniques revealed interactions between the identified umami peptides and T1R1/T1R3 umami receptors as well as TAS2R14 bitter receptors, highlighting critical binding residues. Notably, LT-5 demonstrated a significant capacity to reduce quinine concentration, thereby enhancing the bitter-masking effect and elevating the bitterness threshold. These findings highlighted the potential of tuna-derived peptides as natural umami enhancers and bitterness inhibitors, providing valuable prospects for flavor improvement in food products.