Berlin, Berlin, Germany
Hi, I’m Jeroen - architect, software developer, and founder of PlanFinder. My work sits at the intersection of architecture, machine learning, and computational geometry. I research and build methods for generating, analysing, and reasoning about architectural floor plans and CAD/BIM data. With PlanFinder, I develop AI-assisted tools for early-stage residential floor plan generation and exploration. I also developed Chonk, a geometry-analysis technology for recognising patterns, detecting similarity, and comparing design versions in CAD/BIM models. Across both, I’m interested in how machine learning can help turn architectural geometry into structured design knowledge.
PlanFinder builds Machine-Learning-based software products for the AEC industry. The tools help architects and design teams generate, explore, and analyse architectural design data - from early-stage floor plan generation to clustering and pattern recognition in CAD/BIM models.
Deep Learning on GIS data and research project on combining neural networks with symbolic models (see publication below).
Facade rationalization of Kingdee tower in Shenzhen.