Research

Advancing computational science

Tanuki is built on peer-reviewed research in computational mechanics, scientific computing, and machine learning. We develop methods that are rigorous enough for academia and accessible enough for industry.

01

AI-driven geometry generation

Describe a component in natural language and receive simulation-ready 3D CAD geometry. Fully parametric, watertight, and mesh-ready.

  • Programmatic solid modelling from conversational input
  • Iterative refinement through follow-up instructions
  • Automatic mesh generation with adaptive resolution
  • Export to standard engineering formats

02

High-fidelity simulation

Research-grade finite-element solvers handle structural, thermal, and fluid problems with the rigour expected in academic and industrial settings.

  • Linear and nonlinear elasticity
  • Heat conduction and convection
  • Incompressible flow and shallow water equations
  • Modal analysis and eigenvalue problems

03

Machine learning for physics

Hybrid approaches that combine data-driven models with physical laws, accelerating solvers, learning constitutive relations, and enabling surrogate modelling.

  • Physics-informed neural network integration
  • Surrogate model training from simulation data
  • Adjoint-based sensitivity and shape optimisation
  • Automated model selection and hyperparameter tuning

04

Geospatial modelling

Build ocean meshes, terrain models, and geographic simulation domains from a place name. Real-world data is incorporated automatically.

  • Coastline extraction and domain construction
  • Seafloor bathymetry integration
  • Satellite-derived ice edge boundaries
  • Multi-resolution unstructured meshing for large domains

Academic collaborations

Imperial College London

Finite-element methods for fracture mechanics and deformation modelling, and AI for scientific computing through the I-X Centre.

University College London

Geomechanics, fault mechanics, and fluid–rock interaction research applied to subsurface and environmental systems.

University of Leicester

PDE-constrained shape optimisation and inverse-problem methodologies for geoscience applications.

University of Cambridge

Collaboration on the ARIA Re-thickening Arctic Sea Ice programme, applying computational mechanics to climate intervention research.

If you are a researcher or institution interested in collaborating with Tanuki, we would love to hear from you.

Get in touch