GLOBAL INTENT COMPANY

Learned molecular reasoning architecture that maintains and updates competing molecular hypotheses from mass-spectrometry evidence rather than one-shot retrieval.

Computational architecture organized around selective local processing, bounded global integration, persistent hypotheses, and learned gating.

A small decoder-only Transformer initialized, trained, and executed for inference entirely client-side inside the browser. Written in native JavaScript ES modules and WebGPU WGSL compute shaders with CPU fallback, featuring a custom reverse-mode autograd tensor engine, BPE tokenizer, Adam/AdamW optimizers, and non-blocking Web Worker execution with zero server dependencies, Python runtimes, or build steps.

An experimental pure C Transformer training and inference engine built for Apple Silicon. Explores 3.45M and 23M parameter architectures, subword BPE tokenization designed to address character-level sequence expansion, pure C conversational instruction fine-tuning with prompt-loss masking, pre-allocated activation buffer inspection via a live terminal REPL, and the Rung 6 V2 reasoning dataset curation workflow.

Explores computational hemodynamic modeling and Optic-Trigeminal physiological systems with transparent preservation of null and mixed experimental results.