Developing mechanically resilient liquid-repellent surfaces has gained considerable interest for applications such as self-cleaning coatings, anti-fouling layers, wearable electronics, and biomedical devices. However, achieving long-term durability without sacrificing liquid repellency remains a significant challenge due to the fragile nature of the micro- and nanoscale features required to maintain low solid–liquid contact area. This dissertation systematically addresses this challenge through the design of advanced surface architectures and scalable fabrication methods. First, a hybrid structure combining soft micropillars with rigid doubly re-entrant caps was developed, effectively balancing compliance and rigidity. This configuration preserved omniphobicity while withstanding large mechanical deformations, including compression, bending, and cutting. Building on this, the study introduced purely soft, monolithic doubly re-entrant microstructures, geometrically optimized to avoid collapse and stiction. These structures achieved robust liquid repellency without requiring rigid elements or complex alignment, simplifying fabrication and enabling use on flexible or curved substrates. To further enhance robustness and manufacturing compatibility, the doubly re-entrant geometries were replicated into thermosetting polymers using soft molds guided by theoretical collapse models. This scalable approach enabled high-fidelity replication across large areas, producing structures that retained omniphobicity while exhibiting superior resistance to abrasive and cutting forces. Finally, inspired by springtail skin, a bio-mimetic hierarchical structure with solid fraction gradients was engineered by laminating micropillar films onto wavy substrates. This simple, alignment-free process produced directional, multiscale topographies that pave the way for advanced wetting control. Overall, this dissertation presents a unified framework for designing and fabricating mechanically resilient, omniphobic surfaces, offering scalable solutions for future deployment in real-world applications.