Zer0pa is a falsification-governed, AI-native research and development lab with a public proof surface.

We sit between an applied research lab, an IP foundry, a proof-infrastructure company, and a frontier lab. The work spans motion, music, markets, biosignals, neural events, ancient scripts, vessel design, in-silico science pipelines, and deterministic computation. Across that portfolio, the operating question is consistent: what can be bounded, encoded, tested, broken, replayed, and stated without hiding the limits?

We direct frontier AI systems as researchers, coders, scientific operators, and adversarial thinkers across public repositories. The aim is not one universal platform. The aim is a portfolio of bounded systems that connect scientific corridors: physics, code, falsification, and nature's design patterns turned into testable artifacts.

We are the code.

Zer0pa looks for forms nature already uses: rhythm, contour, motion, branching, recurrence, vessel pressure, molecular pathway, symbol shape, field behavior. The work is to translate those forms into bounded computational artifacts that can be inspected, tested, broken, replayed, extended, and improved without hiding the limits.

Across this portfolio, rigor governs the work. Motion, music, markets, biosignals, neural events, ancient scripts, vessel design, materials, energy, and life-science pipelines are separate domains; the shared discipline is falsification.

We are not conventional open source. We are open for builders: generous where access creates opportunity, disciplined where scale creates monopoly.

Most public surfaces are designed so indie developers, emerging ML engineers, AI builders, and scientific operators can inspect the claim, run the surface that is actually released, and build value before they need a licensing conversation.

Every public product or research front door should land in one of three states.

When in doubt, we classify a repo as RESEARCH-READY. No current public repo is classified as ENTERPRISE-READY.

Zer0pa repos are front doors, not pitch decks.

• Claims are bounded to the repo and its evidence surface.
• Negative results, blocked gates, stale packages, platform limits, and non-claims are first-class outputs.
• Product-readiness claims are kept separate from package-install mechanics.
• Public proof surfaces do not grant private code, datasets, weights, CAD, raw CFD, partner material, manufacturing authority, or internal runtime systems.
• The portfolio is built to be falsified: physics, code, receipts, and stated blockers before narrative.

Deterministic representations for one bounded input domain at a time.

Deterministic computation, falsification infrastructure, and ancient-script research surfaces.

Applied research pipelines where the proof surface names what has been run and what remains open.

Public ship-scale engineering surfaces. These repos publish architecture, authority state, and boundaries without releasing internal CAD, raw CFD, partner data, manufacturing authority, or classification approval.

Most public engineering repos use Zer0pa source-available licensing or repo-specific research licenses. Some methods repos use permissive open-source licensing where the repository states that explicitly.

Canonical legal endpoint: ZPE-License-Commercial.

Public READMEs are not a blanket license grant. Each repo's own license, notices, non-claims, package status, and proof boundaries govern that repo.

• We do not claim one universal algorithm across the portfolio.
• We do not claim production medical, regulatory, class, shipyard, manufacturing, or enterprise approval from research repos.
• We do not hide adverse results below the fold.
• We do not treat a package install as product readiness.
• We do not treat "live on GitHub" as a readiness category.

Commercial licensing, notices, and serious technical discussion: architects@zer0pa.ai

Website: zer0pa.ai