We leverage cutting-edge research in geometry and topology to build models that respect the internal structures of your data
Get in touchNeural architectures for non-Euclidean data. Graph neural networks, manifold learning, equivariant networks.
Persistent homology and topological methods for drug discovery, materials science, and anomaly detection.
Riemannian optimization, natural gradients, and geometric insights for training neural networks.
From research papers to production code. Custom solutions in PyTorch, JAX, and specialized geometric ML libraries.
Mathematical evaluation of AI approaches for investors, research teams, and technical leadership.
Workshops and lectures on geometric methods in machine learning, tailored to your team and domain.