Company Performance Metrics
TensorType develops business software and conducts technical research focused on structural support within deep learning systems. The company creates a pure functional tensor processing framework implemented in the dependently typed language Idris 2. Its framework extends traditional rectangular tensors to support abstract syntax trees, inductive
types, recursive datasets, and multi-step interaction protocols. TensorType emphasizes type safety by construction at compile time, enabling safer experimentation with complex neural network architectures. The framework offers first-class named axes, precise indexing, and structural tensor handling with syntax inspired by PyTorch and NumPy. It is designed for software developers and researchers who prototype structural neural networks and require reliable interaction with complex structural logic.
