Interpretable, physics-informed machine learning is a central theme of my research, aiming to bridge multi-scale, multi-physics understanding with data-driven prediction for catalytic systems. By embedding physical laws, descriptors, and constraints derived from density functional theory (DFT), microkinetic modeling, and transport phenomena into machine learning frameworks, our approach ensures both predictive accuracy and mechanistic interpretability. Rather than treating models as black boxes, we leverage techniques such as feature attribution, symbolic regression, and physically meaningful embeddings to uncover governing relationships between catalyst structure, reaction pathways, and performance metrics. This enables identification of key descriptors, rate-limiting steps, and design rules that generalize across materials and reaction environments. The resulting models not only accelerate catalyst discovery but also provide actionable insights that can be directly integrated into multiscale workflows, including reactor design and process optimization, thereby supporting reliable, explainable, and scalable innovation in catalysis.