Machine Builder · Self-Taught AI Researcher (since 2026) · AI-Assisted Software Development
Vincit Omnia Veritas — Truth Conquers All
$ whoami
Chad Holland
Machine Builder | Self-Taught AI Researcher (since February 2026)
I build and investigate experimental AI systems, evidence verification,
offline memory systems, and reproducible research methods.
I direct research and development using several AI models for drafting,
implementation, critique, and testing. AI agreement is not independent
validation. Failed tests, open questions, and limitations remain visible.
My background is in industrial machine building. I began exploring AI research in February 2026 and remain a self-taught beginner in the field. I use multiple AI systems under human direction to develop and challenge experimental software. A passing test or AI-generated review is evidence of a bounded check, not proof of general correctness or independent security validation.
- Hugging Face · @holland202 — public model and dataset repositories.
- LDE Cross-Substrate Research dataset — experimental results and evidence.
- Sovereign Evidence Bench dataset — benchmark artifacts and reproducibility material.
- LDE Cross-Substrate Research model repository — a research repository, not a claim of a validated foundation model.
- LinkedIn — professional background and research updates.
- ORCID · 0009-0004-1943-0585 — persistent public researcher identifier; not a certification of research findings.
- Local Discovery Engine on Codeberg — project source and experiments.
Vincit Omnia Veritas — Truth Conquers All.
Statistics are supplied by third-party services. Contribution totals may differ from public-repository activity depending on GitHub visibility settings and provider methodology.
How the percentages are calculated
By default the card ranks languages by byte count of source files across the user’s non-forked repositories (size_weight=1, count_weight=0). It is a GitHub metric of code volume, not a measure of personal proficiency, and not one primary-language vote per repository. Public-instance results reflect public repositories.
(historical snapshot: 22 October 2026 — exclusive assignment, no double-counting; not a live repository count)
| Category | Repositories | Share |
|---|---|---|
| Principia / Research notes | 1 | 3.8 % |
| Veritas / Evidence & Gate | 9 | 34.6 % |
| Geometric / Learning | 2 | 7.7 % |
| Edge / Runtime / Swarm | 4 | 15.4 % |
| Tools / Other | 10 | 38.5 % |
| Total | 26 | 100 % |
Representative members:
- Principia →
principia-artificialis - Veritas →
sovereign-veritas,veritas-science,veritas-holo,eunoia,token-veritas,veritas-companion,vacuity_lint.py,evidence-ledger,veritas-eval-harness - Geometric →
quasar,quasar-v2 - Edge / Swarm →
skn-v1-,slc-v12-,sovereign-evolution,sovereign-suite - Tools / Other → remaining public repositories
Generated daily in this repository by
scripts/activity_graph.pyfrom GitHub's public contribution calendar. The image states the date of its last data point.
Updated daily with
yoshi389111/github-profile-3d-contrib. Uses a token scoped to this public profile repository; private repositories are not accessed. Updates stop if GitHub reports restricted contribution counts. Generated images live on a separate branch.
Defect reports and pull requests on repositories I don't own. Historical status snapshot from 2026-10-22; follow links for current state.
| Project | Contribution | Status | Link |
|---|---|---|---|
| opena2a-standards/aap-conformance | Showed a verifier that skips the presenter-proof signature check (§4.6) still passed all 43 conformance fixtures | Fixed by the maintainer (PR #6, commit f0d1339) |
Issue #4 |
| opena2a-standards/aap-conformance | Two fixtures pinning the UNKNOWN_KEY checks (follow-up to #4) | Open pull request, not yet reviewed | PR #8 |
| opena2a-org/oasb | Most test files bypass the adapter, so only 69 of 245 tests reach it | Open | Issue #40 |
| GenAI-Security-Project/agent-control-standard | toolCallRequest has no idempotency binding, so a retry can repeat an external effect |
Open | Issue #200 |
| johnbenac/graphmd | Clarify slash-ref ambiguity for GH-005 repository imports | Closed, not merged | PR #438 |
| Project | Description | Status | Link |
|---|---|---|---|
| Principia Artificialis | Open research program on the mathematics of artificial thought. Notes with registered predictions; failures retained. | Exploratory · public dashboard live | Repo · Dashboard |
| Sovereign Veritas | Fail-closed permission gate for AI actions. Decisions packageable, signed, independently re-verifiable. Built and run on a phone. | Experimental · not production-ready | Repo |
| Sovereign Memory Layer (SML) | Offline evidence-aware memory store with Python, Rust, and TypeScript implementations and differential tests. | Experimental · documented outside adversarial testing · no human code audit | Repo · Testing |
| veritas-science | Executable scientific validation framework for adversarial, preregistered, evidence-first experimentation. | Public research tooling | Repo |
| veritas-holo | Falsifiable geometric reasoning substrate: operators, invariants and holonomy, measured before believed. | Public research | Repo |
| eunoia | Evidence-bounded architecture for autonomous intelligence. | Public research · not production-ready | Repo |
| quasar | Closed-loop learner that generates its own training data. | Public research | Repo |
| skn-v1- | Topology-aware swarm robotics with information geometry and post-quantum trust. | Active research framework · partial implementation | Repo |
| slc-v12- | Hardware-bound, deterministic AI runtime for edge devices featuring thermal-aware manifold inference and Veritas Gate governance. | Research / experimental | Repo |
| Project | Public status | Independent reproduction | Key public limitations | Evidence |
|---|---|---|---|---|
| Principia Artificialis | Exploratory notes program. “Verified” = reference code prints claimed numbers. | Bounded external checks recorded; no broad scientific replication claimed. | AI assistance disclosed; many notes Speculative or Draft. | Repo · METHOD.md · Dashboard |
| Sovereign Veritas | Experimental fail-closed gate. | Bounded external reproduction of demo + contract recorded. | Not production-ready · not independently security-validated. | Repo · STATUS.md · Issue #79 |
| Sovereign Memory Layer | Offline memory store; Python, Rust, and TypeScript ports, with tests documented in-repo. | One outside preregistered agent-executed adversarial test and outside AI-assisted code review reported; not a human-run audit. | Retrieved values are untrusted; KNOWN does not establish truth. Real power-loss and hostile-user claims untested. | README · BABEL-001 |
| skn-v1- / slc-v12- | Research frameworks with measured subsystems and explicit design-vs-implemented boundaries. | None claimed beyond repository tests. | Documented in each README. | skn-v1- · slc-v12- |
Primary public entry point
→ https://holland202.github.io/principia-artificialis/dashboard.html
Live status counts, evidence cards, open questions, and contribution entry points. Generated from the public repository and checked in CI.
huggingface.co/holland202 is the public companion to this GitHub account: LDE research evidence, Sovereign Evidence Bench, and LDE model repository. These are experimental artifacts, not endorsements or independently validated model-performance claims.
Open to ideas beyond my existing projects. If you have something you want to build, investigate, test, document, or improve together, you're welcome to propose it—even if it has nothing to do with my current repositories.
People from any background or experience level are welcome: developers, researchers, engineers, designers, writers, students, independent builders, and curious newcomers. Possibilities include new software, original experiments, reproducibility checks, adversarial review, documentation, accessibility, or a completely different idea you bring.
- Bring your own proposal. You do not have to join one of my projects. Explain the goal, what you hope to contribute, and what kind of collaboration would help.
- Work on an existing project. Inspect code, challenge hypotheses, reproduce results, report defects, or help with documentation and testing.
- Set the scope together. Participation is voluntary and unpaid, at your own pace, with no obligation or fixed schedule. We can decide whether a project is a good fit before starting.
How to reach out: For an existing public project, open an issue in its repository. For research-related ideas, start with Principia Artificialis issues, or use an existing public contact channel to share a proposal.
Helpful starting points: Research dashboard · Principia Artificialis issues · Sovereign Veritas collaboration invitation
Critical review, contradictory evidence, and negative results are welcome. Collaboration proposals are invitations to discuss, not automatic approval to modify a repository or run experiments. This profile does not list or link private repositories.
Chad Holland · Machine Builder · Independent AI Researcher · United States
Language percentages reflect byte-count ranking of public repositories.
AI assistance in research and development is disclosed in the relevant repositories.





