Multimetallic nanoparticles are of interest as functional materials due to their highly tunable properties. However, synthesizing congruent mixtures of immiscible components is limited by the need for high-temperature procedures followed by rapid quenching that lack size and shape control. Here we report a low-temperature (≤80°C) non-equilibrium synthesis of nanosurface alloys (NSAs) with tunable size, shape and composition regardless of miscibility. We show the generality of our method by producing both bulk miscible and immiscible monodisperse anisotropic Cu-based NSAs of up to three components. We demonstrate our synthesis as a screening platform to investigate the effects of crystal facet and elemental composition by testing tetrahedral, cubic and truncated-octahedral NSAs as catalysts in the electroreduction of CO2. The use of machine learning has enabled the prediction and informed synthesis of both multicarbon-product-selective and phase-stable Cu-Ag-Pd compositions. This combination of non-equilibrium synthesis and theory-guided candidate selection is expected to accelerate test-learn-repeat cycles of structure-performance optimization processes.