Skip to main content
  1. Home
  2. Research

Independent research

Research across intelligent systems, mathematics, and future compute.

My research asks how AI systems can reason across fragmented information, preserve useful context, coordinate actions, and remain accountable to the people who use them.

I also study longer-horizon questions in mathematical foundations and compute architecture. The work is separated below by evidence status so production findings, applied R&D, and exploratory hypotheses remain distinct.

How to read this page

Every item carries an evidence status.

The distinction is part of the research, not a disclaimer added afterward. Nothing here is described as peer reviewed.

Production validated
Real deployment data, a disclosed methodology, reviewable evidence, and a direct source.
Research preprint
Work posted to arXiv. A preprint has not been through journal peer review, and is never described here as if it had.
Applied R&D
Implemented systems, simulations, experiments, or working technical notes that are not yet a formal publication.
Architecture essay
A reasoned technical proposal without empirical validation.
Exploratory hypothesis
Long-horizon theoretical work where the claims are not empirically established.

Production research

Evidence from a live deployment

Research preprint Production data

Decision Traces: What Multi-System Data Fusion Reveals About Institutional Knowledge in Enterprise Hiring

Enterprise hiring data was distributed across an applicant tracking system, a behavioral assessment, and production outcomes. Connecting those systems produced evidence that none could reveal independently.

Dataset. 10,765 hires from 2022–2025 at a Fortune 500 insurance carrier.

Selected findings. Of 8,181 unique skills parsed from ATS profiles, 3,597 had enough data to test. 0 predicted sustained production after Bonferroni correction, and 30 were significantly anti-predictive. The behavioral assessment achieved AUC 0.647 alone; multi-system fusion reached AUC 0.735.

Speed to production carried a measurable economic value of approximately $54 per agent per day unadjusted, or $35 per day after controlling for source channel and tenure.

Full research page arXiv abstract PDF BibTeX Nodes evidence and methodology

Applied systems research

Questions I am actively working on, and what I have written about each.

These are research areas, not finished results. Where I have published working notes on a question, the note is linked. Nothing below claims a solved result.

Applied R&D

Mathematical foundations of memory and reasoning

Ongoing work on how state, memory, representation, uncertainty, optimization, and learning dynamics affect persistent intelligent systems.

  • Working notes, not yet public

Future compute

Future compute architectures

Long-horizon exploratory work on architectures that may overcome limits in conventional AI compute, including photonic, brain-inspired, and energy-based approaches. These are architecture proposals and hypotheses, not production claims.

Architecture essay · 2025

Towards True ASI: A New Paradigm in Hardware and Brain-Inspired Architecture

Explores limitations in current silicon-centric AI systems and proposes a direction combining photonic or energy-based computation with specialized, dynamically activated subsystems inspired by biological organization.

Medium
Exploratory essay · 2025

Quantum, Meta-Computational, and Cosmic Approaches to Advancing AI

Examines possible connections between alternative computation, quantum-scale information, and long-horizon AI architecture.

Medium
Exploratory hypothesis · 2025

Quantum-Cosmic Meta-Intelligence

Proposes a speculative framework connecting intelligence, information, energy, and physical reality. Presented as a hypothesis and conceptual synthesis, not an empirical result.

Academia.edu · self-archived, not a peer-review venue
Exploratory hypothesis · 2024

Quantum Divine Substrate Theory

A speculative metaphysical and physical model proposing a common substrate beneath matter, consciousness, and space-time. It is not established quantum physics and is not presented as such.

Academia.edu · self-archived, not a peer-review venue

Research standard

How I label certainty

I distinguish between evidence, implementation, and hypothesis. Production findings include the dataset and method. Applied R&D identifies what was built or tested. Architecture essays identify what is proposed. Exploratory work is labeled as such.

That distinction is part of the research, not a disclaimer added afterward.

Sources and indexes. arXiv hosts preprints. Google Scholar and Academia.edu are indexes and self-archiving profiles, not publication venues. Where a piece has not been through journal peer review, this site says so.

Research collaboration

For research collaboration, replication questions, technical discussion, or speaking, get in touch with the relevant paper or research area in the subject line.

Email about research All contact routes