Inhibiting the programmed death-1 (PD-1)/programmed death ligand 1 (PD-L1) axis by monoclonal antibodies (mAbs) is a successful cancer immunotherapy. However, mAb-based drugs have various disadvantages including high production costs and large molecular sizes, which motivated us to develop a smaller alternative drug. Since PD-L1 binds PD-1 with moderate affinity, a higher affinity PD-1 variant should serve as a competitive inhibitor of the wild-type PD-1/PD-L1 interaction. In this report, we conducted in silico point mutagenesis of PD-1 to identify potent PD-1 variants with a higher affinity toward PD-L1 and refined the in silico results using a luciferase-based in-cell protein-protein interaction (PPI) assay. As a result, a PD-1 variant was developed that had two mutated amino acids (T76Y, A132V), termed 2-PD-1. 2-PD-1 could bind with PD-L1 at a dissociation constant of 12.74 nM. Moreover, 2-PD-1 successfully inhibited the PD-1/PD-L1 interaction with a half maximal inhibitory concentration of 19.15 nM and reactivated the T cell with a half maximal effective concentration of 136.1 nM. These results show that in silico mutagenesis combined with an in-cell PPI assay verification strategy successfully prepared a non-IgG inhibitor of the PD-1/PD-L1 interaction.