Zain Dana Harper

PROJECT TELOS

A place to see and make things — together with a model.

I build tools where a person and an AI look at the same thing, talk about it, and shape it together — a drawing, a photo, a piece of code, an argument. Everything they make is yours to open up and check, but that’s the floor. The point is what you can build on it.

Try the Studio → See the work Read the thinking
Macrophotography by iamBeTa — the living structures these ideas are modeled on, redrawn by algorithm in the atelier.

Wherever someone hands work to a model — or hands a model's work to someone else — the same gap opens: you can't see what it saw, and you can't check what it claims. The same gap, in a clinic, a courtroom, a newsroom, an audit, a repo. Here is what closing it looks like.

Radiology.

Without the radiologist sees a label, not what the model saw → Trust by reputation; the error surfaces downstream, if ever.
With clinician and model stand before the same witnessed frame. → A second read you can actually interrogate.

Your codebase.

Without a silent overwrite, an unrun test, a fabricated green check — you find it in production. → Debugging a failure the agent said couldn't happen.
With the work is fingerprinted before and after; you see exactly what changed and why, re-derivable. → You hand a model real work and still hold the receipts.

Newsroom.

Without the clip is a re-encode of something older; both publish as authentic and the correction comes after it spreads. → A retraction, and eroded trust.
With origin is witnessed (content + perceptual); drift is flagged before it ships. → You publish what you can vouch for.

See how this plays out across fields →

Zain Dana Harper.

One rule runs under every organ above: build it, mark where it stops, and leave the evidence out where you can check it. I wrote the thesis on this page by trying to embarrass myself — going at my own claims the way an attacker would — and it keeps a record of the three times I got through. EMET won’t say “trusted.” The roadmaps mark what shipped, what’s still an experiment, and what’s only an idea. That isn’t modesty. It’s the work refusing to flatter me.

The honest version of who’s behind it: I’m not really a programmer. I think in systems, and fast — a brainstormer who throws off more ideas than he can keep track of, ideas that won’t sit still or keep themselves honest on their own. So the first thing this harness ever held was me; I built the discipline I needed for my own head, and only later noticed it worked on everything else. There’s no computer-science degree behind any of it — my path ran through eleven years of commercial arboriculture in the family business, a contract diagnosing Xbox Live networking, then a long, largely private sprint across compilers, real-time graphics, color science, security, and AI accountability, the offensive side and the defensive one, because you learn how a system holds by learning how it breaks. With no institution to vouch for me, I learned early to let the work answer for itself or go unread. Independent since 2023, out of Seattle.

The closest title is forward-deployed systems architect — turned around: not embedding with a customer to build on a platform that’s already there, but going onto the platform that isn’t there yet — the accountable layer the rest of this will one day stand on.

One standard runs under all of it, and it’s the one I’m asking you to hold me to: judge the action, never the worth of who took it. That is the only thing that lets a thing be wrong without being worthless — the only thing that lets the fool become the scholar.

That’s the short version. The longer one — how this actually came to be, who it’s really for, and the person still working it out — is its own page now, kept apart on purpose so the work never leans on it. The person behind it →

Contact.

For collaboration, evaluation, or the long argument — reach me directly. No funnels; no forms pretending to be people.

Email — zaindharper@gmail.com  ·  GitHub — HarperZ9  ·  CV — résumé · full CV  ·  Seattle · remote-ideal