As we navigate the aftermath of the COVID-19 pandemic, the call for advanced medical countermeasures to preempt the next pandemic has echoed across the realm of drug development. Within the pandemic toolbox, drugs and vaccines have been demonstrated as invaluable agents to curb the impact and disease burden of widely disseminated infection, though discovery of novel therapeutic agents remains a major bottleneck. The advent of deep learning-guided protein design offers promise in rationally designing novel, broadly neutralizing therapeutic agents that can be useful for pandemic scenarios. Herein, I apply these deep learning tools to develop miniproteins that cross-react and neutralize viruses with pandemic potential. As a proof of concept, I focus on the henipaviruses (namely Nipah and Hendra virus), a family of zoonotic viruses that top biosecurity watchlists because of their propensity to infect via a variety of routes, their difficulty in diagnosis, and their lack of approved treatments - drugs or vaccines. I also use this platform to identify strategies to improve drug-like properties of proteins in low/medium-throughput assays. Further, I develop and optimize a high-throughput screening modality that tags protein library members with short peptide barcodes designed for readout on high resolution mass spectrometry. I then use this method to screen de novo protein 3libraries for soluble expression, designed assembly state, and reactivity with their designed target. Looking forward, this work provides a framework for developing drug-like miniproteins that cross-react with multiple homologs using known structural & functional information. Moreover, this peptide barcoding screen offers a streamlined approach for multi-parameter evaluation of entire protein libraries. Combined with in silico design and prediction, such an assay could be useful for informing computational tools to improve future rounds of computational design.