EngiBench: A Framework for Data-Driven Engineering Design Research

Florian Felten (ETH Zürich) · Gabriel Apaza (University of Maryland, College Park) · Gerhard Bräunlich (ETHZ - ETH Zurich) · Cashen Diniz (ETHZ - ETH Zurich) · Xuliang Dong (University of Maryland, College Park) · Arthur Drake (University of Maryland, College Park) · Milad Habibi (University of Maryland, College Park) · Nathaniel Hoffman (University of Maryland, College Park) · Matthew Keeler (ETHZ - ETH Zurich) · Soheyl Massoudi (ETHZ - ETH Zurich) · Francis VanGessel (University of Maryland, College Park) · Mark Fuge (ETHZ - ETH Zurich)
benchmarksdata-driven engineering designdataset generationdesign manifoldsend-to-end experiment workflowsengineering design optimizationfeasibility checksgenerative modelsmachine learning algorithmsmodular librariesopen-source libraryperformance analysisphysics-based simulationssurrogate modelsunified api

Engineering design optimization seeks to automatically determine the shapes, topologies, or parameters of components that maximize performance under given conditions. This process often depends on physics-based simulations, which are difficult to install, computationally expensive, and require domain-specific expertise. To mitigate these challenges, we introduce EngiBench, the first open‐source library and datasets spanning diverse domains for data‐driven engineering design. EngiBench provides a unified API and a curated set of benchmarks