We employed PPIFlow (Yu et al., bioRxiv 2026), a flow-matching-based generative framework for de novo protein binder design. PPIFlow uses a pairformer architecture to model protein backbone rigid-body transformations as continuous flows, conditioning on hotspot residues at the target interface. Generated backbones were passed through AbMPNN for VHH sequence design, followed by AF3Score filtering to prioritize high-confidence candidates. Selected candidates then underwent in silico affinity maturation via interface rotamer enrichment combined with partial flow refinement to optimize interfacial packing. Final candidates were validated with template-free AlphaFold3 structure prediction.