De novo designed binder for the Nipah virus attachment glycoprotein aiming to occlude the ephrinB2 binding site. Trajectory was generated using bindcraft, validated by boltz2 using the ipsae metric then optimized for solubility and ipsae score increase by 2 rounds of MPNN using soluble weights. In a way this approach was largely competition-centric, as I would usually stick to AlphaFold2 iptm values otherwise. This particular fold-interface was selected as it is predicted with boltz2 consistently well across iterations and is a good compromise between size and epitope occlusion.