UpSkillZone

Public manifesto · v1.0 · 2026-04-12

We sell outcomes, not lectures.

A short statement of what we believe, what we will issue, and what we refuse to issue. Read once; cite forever.



§1

What we believe

Traditional certificates fail because they assert without proving. A PDF that says "completed Module 4" is a claim, not evidence. The hiring manager reading it has no way to check the rubric, the grader, or the artifact behind the claim. So they discount it to zero, and the candidate carries a credential the market has already priced as noise.

Verifiable credentials with calibrated mentor review solve this. Each skill assertion ships with a signed artifact, the identity of the mentor who graded it, and the inter-rater kappa that mentor holds against the calibration set. The credential is a W3C VC anyone can fetch from https://upskillzone.com/cert/... and verify against the issuer JWKS. No platform mediation, no login wall, no "trust us".

The moat is not the curriculum. The moat is the public outcome ledger: every credential, every grader kappa, every 90-day outcome attestation, queryable by employers at the skill level. Once the ledger is large enough that hiring managers reach for it instead of LinkedIn, the credential becomes the default — and the schema we use becomes the standard.


§2 · The seven commitments

How that translates into product

I · Commitment

Verifiable credentials

Every Job Twin pass produces a W3C VC + Open Badges 3.0 assertion, Ed25519-signed by the issuer key, subject DID written as did:web:upskillzone.com:u:{handle}.

II · Commitment

Calibrated mentor review

Every grader holds a Cohen's kappa above 0.7 against the calibration set before they touch a learner artifact. Drift below the floor pulls them out of rotation automatically.

III · Commitment

Public skills directory

Opted-in talents appear at /skills indexed by skill key. Employers query the ledger, not us.

IV · Commitment

Ed25519 hire attestation

When an employer hires a verified talent, they sign a short attestation referencing the credential URL. The hire event becomes a first-class object in the ledger.

V · Commitment

90-day outcome attestation

At day 90, the hiring manager signs a second attestation on whether the talent shipped what the credential predicted they could. This is the feedback loop legacy EdTech never built.

VI · Commitment

Outcome credit refund

If a track does not place a learner against its stated outcome within the published window, the outcome credit is refunded. The clause is in the enrollment agreement, not a footnote.


§3 · VII

What this rules out

  • No resume-style certificates. A line item with no signed artifact behind it is a claim, and we do not issue claims.
  • No peer reviews as the primary signal. Peer review is useful as a second pass; it is not a substitute for a calibrated mentor with a kappa on file.
  • No auto-grading-only credentials. A regex-graded answer tells you whether the regex matched, not whether the learner can ship the work.
  • No opaque grading rubrics. Every Job Twin rubric is public before the learner attempts the artifact, and the grader comments are attached to the credential.
  • No proprietary credential schema. We issue against W3C VC Data Model 2.0 and Open Badges 3.0 so the credential survives us — if we wind down, the credentials still verify.

Next

Read the ledger, not the pitch