The use of peptidomimetic scaffolds to target protein protein interfaces is a promising strategy for inhibitor design. The strategy relies on mimicry of protein motifs that exhibit a concentration of native hot spot residues. To address this constraint, we present a pocket-centric computational design strategy guided by AlphaSpace to identify high-quality pockets near the peptidomimetic motif that are both targetable and unoccupied. Alpha-clusters serve as a spatial representation of pocket space and are used to guide the selection of natural and non-natural amino acid mutations to design inhibitors that optimize pocket occupation across the interface. We tested the strategy against a challenging protein protein interaction target, KIX/MLL, by optimizing a single helical motif within MLL to compete against the full-length wild-type MLL sequence. Molecular dynamics simulation and experimental fluorescence polarization assays are used to verify the efficacy of the optimized peptide sequence.