About the role
About Us Vinci is building physics AI for hardware design. Our models deliver ultrafast, accurate thermal and mechanical simulation and our platform puts that capability in the hands of every hardware engineer, not just simulation specialists. The quality of our models is bounded by the quality and diversity of the data they're trained on, and that data doesn't exist in the wild. We generate it. General Description You will join the data generation team and build the systems that produce the synthetic geometries, materials, boundary conditions, and simulation configurations used to train and evaluate our physics models. This is a deep technical role at the intersection of computational geometry, physics simulation, and machine learning. The core challenge: write programs that generate families of hardware-like geometries and simulation setups — diverse enough to cover the space our models will see in production, constrained enough that every sample is valid, physically plausible, and useful as training signal.
What you'll do Design and build procedural generators for parametric, hardware-like geometry using programmatic CAD (e.g., CadQuery, OpenCascade, Build123d) or another tool of your choice. Generate physically plausible simulation configurations — boundary conditions, material assignments, loading scenarios — that respect real engineering constraints. Define and measure diversity and coverage of generated distributions, and close the loop between dataset composition and model performance.
Qualifications Software engineering skills, especially in Python Hands-on experience with programmatic/parametric geometry: scripted CAD, B-rep and mesh representations, SDFs, or procedural generation in graphics tools. Solid geometry processing fundamentals: meshing, boolean operations, voxelization/rasterization, and their numerical pitfalls. Comfort reasoning about physical validity — you don't need to be a simulation expert, but you should care whether a boundary condition makes sense.
Nice to have Exposure to ML training pipelines and how dataset composition drives model behavior Familiarity with LLM-driven code generation. FEA or thermal engineering experience. Familiarity with quality-diversity algorithms, open endedness, deep learning fundamentals. Familiarity with the hardware or semiconductor design process. We're hiring across levels from strong mid-level through staff.
