> Markdown version of [/jobs/ext/4593802-founding-engineer](https://www.wearedevelopers.com/jobs/ext/4593802-founding-engineer). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md). --- # Founding Engineer - **Company:** 1zu100 - **Location:** München, Germany - **Contract:** Permanent contract - **Skills:** Computer-Aided Design, Artificial Intelligence, ArchiCAD, Autodesk Revit, Building Information Modeling, Regression Testing, Reinforcement Learning - **Published:** November 6, 2026 - **Apply:** https://www.careerjet.de/jobad/de536a8bf570c37aeb1e1033a568ed5df9 ## About the Role Hands-on experience with harness engineering: agents, tool orchestration, evals. High agency and an entrepreneurial drive. Ideally, you have founded before or want to. You find the work yourself and move fast. Experience in a high-growth environment (startup, scale-up or similar). Nice to have: German, Revit, IFC or Grasshopper, a non-linear CV and experiments with reinforcement learning (MCTS, GRPO, RLHF) or diffusion models (discrete and continuous). ## Description 1zu100 is automating construction planning. Together with exceptional architecture firms, we plan multi-unit residential buildings and deliver completed, permitted plans as a service. TL;DR We are looking for one cracked founding engineer to build the AI that plans large residential buildings end to end: the agent harness, the data and context layer, and the evals. You work in a small, high-agency team on real projects that ship right away. Your role Harness engineering: You build and run the layer around the model end to end: memory, tool orchestration, verification, drawing and document generation, so the output stays consistent and code-compliant at scale. Data and context: You turn messy planning data (CAD/BIM models, PDFs, site surveys, building codes) into the right context at every phase, always traceable back to its source. Evals and benchmarks: You build the eval harness (datasets, regression tests) and measure our agents against the architectural rule sets and judgment calls. BIM and IFC automation: You connect to the architect's toolchain (Revit, ArchiCAD, IFC) to drive automated modeling and clean interoperability.