Background and Objective: Identifying a milk-clotting enzyme (MCE) with high κ-casein specificity and heat sensitivity remains a challenge in cheese production. Current microbial, plant, and recombinant MCEs often exhibit low clotting activity, poor κ-casein specificity, and high thermostability, compromising cheese quality and increasing production costs. In this study, to address this, we developed a computational pipeline combining structural analysis, machine learning, and molecular dynamics simulation to design approximately 160, 000 peptides from the Rhizomucor miehei protease–Pepstatin A complex (PDB ID: 2RMP). Material and Methods: Single-site mutagenesis, ML-driven affinity re-prediction, and physicochemical filtering yielded 84 peptides. Their specificity as aspartic proteases were validated via predicted Pepstatin A binding and further screened for cross-reactivity with αs1-, αs2-, and β-caseins. Results and Conclusion: Two candidates, Pep1 and Pep2, demonstrated superior κ-casein binding affinities (ΔG =-50. 20 and-39. 07 kcal/mol at 40 °C, respectively), lower melting indices (-4. 05 and-3. 09), and significantly enhanced specificity scores (-10. 85) compared to Rhizomucor miehei protease (ΔG =-33. 9 kcal/mol at 45 °C,melting index = 0. 17,specificity score = 0. 85). These peptides represent promising vegan-and halal-friendly alternatives to chymosins, pending experimental validation. Keywords: Milk-clotting enzymes, Computational peptide design, κ-casein specificity, Thermolabile peptides, Sustainable cheese production, Molecular dynamics simulations, Machine learning, Vegan cheese production, Halal cheese production, Aspartic protease