Projects for The School of AI

A novel AI architecture derived from cellular computation — replacing discrete layers with Hill-function ODEs, MoE gating with receptor-docking affinity routing, and KV cache with fixed-dimension epigenetic trace memory.

DeepSeek‑Dragon – a bio‑inspired neural architecture combining Local Gain Control Units, Predictive Coding Kernels, degree‑preserving sparse routing, and a morphogenetic architecture generator

Quantitative dataset of 119 neural architectures (2017-2025) scored on hardware compatibility and ecosystem friction. Validates the Transformer Attractor thesis.

Neural intelligence silo — Society of Minds architecture with hierarchical memory, 6 specialist SLMs (57.2M params), decision engine bridge, and mesh node distribution. Intelligence to be absorbed by all organization users.

Official Repo of OpenArchX Framework.

🐉 The transformer is a brilliant hack scaled past its limits. DREX is what comes next — tiered memory 🧠, sparse execution ⚡, and a learned controller that knows what to remember 💾✨

DΞCIBΞLION is an audio intelligence module forged in the labs of OBINexus, where noise meets logic and shouting is a feature, not a bug. It mathematically analyzes human vocal input to determine emotional projection through log-scaled loudness evaluation, using a sacred constant: 85 dB.

ORTHOS: A biologically-inspired neural architecture integrating hierarchical Bayesian filters, sparse attention, and structural plasticity for robust state estimation.

Quatrix — Q-Compass Architecture: novel neural architecture replacing attention with value-based navigation. Base for the Quasar model series.

🔬 TECS-L — Consciousness Continuity Engine. 375+ hypotheses, 130+ experiments, PureField theory source.

BioGPT: A biologically-inspired dendritic language model with cortical columns, NMDA temporal memory, and pruning resilience that beats Pythia-160M

The AI-Native Development Engine. 40+ specialized agents & LoRAs, brain-inspired neural architecture, constitutional governance. From prompt to production. Built to liberate artists.

Benchmarking synthetic cognition for emergent self-awareness. Developed by B.E.C.C.A. Labs — 2025.

Modular PyTorch lab for running configurable deep-learning experiments, with ready-to-use data loaders, training pipelines, and metric tracking for both vision and NLP benchmarks.

BIOMIND: A computational architecture integrating Global Workspace, Integrated Information, and Active Inference theories.

🧠 Self-organizing AI agent with dendrite spawning - decomposes complex goals into neurons, spawns parallel sub-tasks, and aggregates results. Features smart parameter extraction, context isolation, and automatic data compaction. Built for Strava analytics with extensible tool system.