What is The SME Standard?

The SME Standard is a professional school that trains credentialed subject-matter experts — lawyers, finance professionals, and other specialists — to do AI training-data work: writing prompts, building grading rubrics, RLHF annotation, and quality-control auditing for AI labs and data vendors.

Why do subject-matter experts need training for AI data work?

AI-data platforms reject more than 80% of expert applicants — usually not for lack of domain knowledge, but because candidates have never learned the work format: self-contained rubric criteria, preference ranking, golden answers, and evaluation conventions. These are teachable skills.

Who is the training for?

Two audiences: individual experts (attorneys, CPAs, CFAs, PhDs, physicians and other credentialed professionals) who want high-value AI training-data work; and companies — AI-data vendors, labs, and enterprises — that need a trained, quality-consistent expert workforce.

The SME Standard — Professional Training for AI-Data Subject-Matter Experts

The Problem

A $15-billion industry runs on an untrained workforce.

Frontier AI labs now pay doctors, lawyers, and finance professionals $100–$500+ an hour to create training data, write evaluation rubrics, and audit model outputs. But the industry trains nobody:

80%+

of expert applicants rejected by platform screens — mostly for craft, not credentials

0

major platforms offering real skills education — onboarding is unpaid assessments and instruction PDFs

65%

of work unpaid on one documented vendor project due to quality rejections — the cost of no training

Figures from public reporting and platform-worker research, 2025–2026: Bloomberg, TechCrunch, Deel State of Global Hiring, GigDrift, platform review data. Sources on request.

The Solution

A professional school, built to the auditor's standard.

The SME Standard teaches the complete craft of expert AI-data work — the same standard used to audit work inside frontier-lab projects — through graded, artifact-first programs:

What we teach · Positive
Writes self-contained rubric criteria that a blind judge can score without seeing the prompt.

What we train out · Negative
Submits "explains the difference between X and Y" criteria that let a model hallucinate and still score full points.

Every student produces real artifacts — prompt packages, rubrics, golden answers, audit reports — and every artifact is graded against published rubrics with written feedback. The assessment method is itself the lesson.

See the full curriculum →

Why Now

The market is moving up-stack — fast.

Generalist annotation is being automated and rate-compressed. Expert evaluation is exploding. The window to enter at the specialist tier is now.

$6B+

combined annualized revenue of the four leading expert-data vendors (mid-2026)

10×

planned growth in specialist "AI tutors" at one major lab — while 500 generalists were laid off

$1B+

annual spend on reinforcement-learning environments discussed by a single frontier lab

283%

growth of "AI trainer" — the fastest-growing cross-border role of 2025

Start with the Founding Cohort →

Who We Serve

Two audiences. One standard.

For Experts · B2C

Individual professionals

Attorneys, CPAs and CFAs, physicians, PhDs, and specialists who want high-value AI training-data work. Cohort-based training, graded practice, a portfolio, and a professional community — with Legal and Finance tracks first.

Explore the programs →

For Companies · B2B

Vendors, labs & enterprises

Turn applicant-pool waste into yield. Private cohorts for your expert bench, onboarding audits, curriculum licensing, and a certified-talent pipeline — measured in approval-rate lift and rework reduction, not completion certificates.

Explore enterprise programs →

Credentials

Taught by people who have graded the work.

The SME Standard was founded and is taught by practitioners — working subject-matter experts and auditors whose experience spans the roles that define quality in this industry:

Task creation

Authoring expert-level prompts and benchmark tasks for frontier-lab evaluation programs — including scenarios rated at the highest complexity tiers, requiring hours of expert work to solve.

Rubric engineering

Building and QA-ing grading rubrics with 100+ weighted criteria to frontier-project standards — self-contained, objectively judgeable, and defensible to external audit.

QC & auditing

Serving as quality-control reviewers and auditors of other experts' packages on RLHF and benchmark programs — verifying every citation, figure, and criterion against primary sources.

SME training

Building the onboarding guides, rubric explainers, and training decks used to bring new domain experts onto live projects — the origin of this school.

Our instructors' project work sits under confidentiality obligations, which we honor completely: our curriculum is built clean-room from public sources and original synthetic examples, and we never trade in any platform's confidential materials or assessment content. That discipline is a feature — it is the same integrity we teach.

Contact

Talk to us.

Experts and companies alike — tell us what you need and we’ll route you to the right program.