The search for next-generation materials increasingly demands approaches that are faster and more predictive than trial-and-error. Our work began by exploring thin film deposition as a high- throughput method to screen refractory high-entropy alloys. Combinatorial sputtering allowed us to create entire libraries of compositions and rapidly probe phase stability and hardness trends. Thin films proved powerful for capturing intrinsic effects, such as the stability of single-phase BCC alloys, but direct comparisons with arc-melted bulk samples exposed their limits. Microstructural evolution—grain coarsening, segregation, and defect formation—often broke the simple link between film hardness and bulk yield strength. These insights motivated us to expand from thin films toward systematic bulk synthesis and validation. To accelerate this process, we have coupled artificial intelligence (AI) and machine learning (ML) with experimental design. Our models predict yield strength and ductility across multi-principal element alloys, narrowing vast compositional spaces to promising candidates. Yet they also reveal their blind spots: defect-driven phenomena like segregation-induced embrittlement remain difficult to capture. I will also introduce our automated laboratory platforms, where alloy synthesis, property measurement, and characterization are integrated into closed-loop, “self-driving” workflows. These systems enable rapid iteration between AI predictions and experiment while building the rich datasets needed to improve future models. Together, thin film screening, bulk validation, and AI-driven automation point toward a new paradigm in materials discovery—where human intuition, machine learning, and autonomous laboratories combine to design structural alloys with unprecedented speed and precision.