The oxygen reduction reaction (ORR) is the limiting bottleneck in alkaline fuel cells, a key hydrogen energy technology. While platinum (Pt) catalysts have historically delivered high activity, they face limitations due to resource scarcity and cost. Consequently, the search for earth-abundant, durable, and high-performance electrocatalysts remains a priority for accelerating the adoption of clean hydrogen energy. This dissertation focuses on investigating spinel oxides as ORR catalysts, a class of materials that combine compositional flexibility, earth-abundant elements, and strong electrochemical stability. I begin by first establishing the fundamental principles and tools by which I will go about characterizing these materials, the chief among them being density functional theory (DFT). Next, I use Co$_3$O$_4$ as a case study to establish a procedure for studying spinel catalysis systematically. Then, by employing DFT at unprecedented scales, I evaluate the ORR across over 300 spinel oxide compositions, spanning both low- and high-entropy regimes. Through approximately 50,000 fully solvated joint density functional theory (JDFT) calculations, I uncovered a transformative mechanism for ORR on the (100) spinel surface: contrary to the conventional view that tetrahedral sites dominate, octahedral sites under oxidized conditions drive the key reaction steps. This finding overturns decades of assumptions and points to new strategies for tuning activity via octahedral cation selection and surface engineering. To further accelerate catalyst discovery, I developed machine learning (ML) models that predict ORR activity from elemental composition, offering rapid screening and partial DFT evaluation methods. These models achieve near-DFT accuracy, significantly reducing the computational effort required for ranking candidate materials. In parallel, I created a fully automated toolchain for data management and visualization, featuring real-time interactive energy level diagrams and dashboards that map the thermodynamics and kinetics of complex reaction pathways. These capabilities minimize manual data processing and enable researchers to explore strain, applied potential, and reaction intermediates dynamically. Crucially, the computational intensity of this work was supported by multiple supercomputing grants that I independently secured, totaling several million dollars in resources. Taken together, my results provide not only a fundamentally revised picture of ORR on spinel oxides but also a scalable computational platform integrating high-throughput DFT, machine learning, and interactive visual analytics. This dissertation thus paves the way for the design and optimization of cost-effective, high-performance catalysts for next-generation hydrogen energy systems.