AI drug discovery on Apple Silicon. De novo generation + drug repurposing. 100% local, no cloud, fully auditable.

Structured State Matrix Architecture (SSMA) is a high-performance framework designed for efficient sequence modeling, combining structured state space models with adaptive attention mechanisms.

Code and data for: Three Phases of Expert Routing — How Load Balance Evolves During MoE Training

Developed a Lasso Regression model applying L1 regularization for housing price prediction. The model achieved an R² score of ~0.64 while performing implicit feature selection and maintaining strong predictive performance.

PERSPECTIVE v2 — A 1.05 trillion parameter sparse Mixture-of-Experts language model that runs on consumer hardware (4 GB VRAM + 32 GB RAM). Features O(1) perspective decay recurrence, 3D torus manifold routing, native ternary {-1,0,+1} weights, holographic distributed memory, and hard geometric safety constraints. Built in Rust.

LiDAR point cloud completion for unstructured terrain in autonomous earthworks