Research projects and open software from AI-Aided Engineering.
CAD-native aerodynamics · Research project
Aerodynamic predictions directly from parametric CAD, without meshing the input geometry. Gradients guide the search for better designs.
19.8% lower surface-pressure error
On HiLiftAeroML, compared with AB-UPT: relative L₂ error 4.53% versus 5.65%.
Probabilistic weather forecasting · Cross-team research
Atlas is a scalable framework for probabilistic medium-range weather forecasting. It learns multiscale atmospheric dynamics using a compact latent representation together with a local projector for high-resolution physical fields, and supports several probabilistic estimators. Atlas was developed through a broad collaboration across NVIDIA.
Open-source software · Operator learning
NeuralOperator is an open-source PyTorch library for developing, training, and evaluating neural operators. It provides model implementations, data processing, training utilities, and examples for learning solution operators of partial differential equations and other physical systems.