Careers at Orion
Built by annotators,
for annotators
We spent years doing this work for platforms that treated us as resources to be deployed. Orion exists to do it right: real careers, transparent pay, and a seat at the cutting edge of AI — as valued employees.
What we promise
Valued employees, not resources to be deployed
A real career ladder
Tasker, Reviewer, Project Lead — a defined path with defined pay at every rung. Advancement is earned through your quality record, not luck or favoritism.
Transparent, above-board pay
Every task shows its rate before you start. Earnings post to your dashboard instantly, paydays are semi-monthly and predictable, and there are no fees, holds, or opaque hourly buckets.
Paid training, always
New skills, new task types, and certifications all happen on the clock. Getting better at this work is part of the job — so we pay for it like it is.
Bonuses & incentives
Quality streaks, calibration accuracy, and project milestones carry real bonuses. When your work makes a project succeed, you share in that.
Flexible hours, competitive pay
Work when it suits you — an evening of comparisons or a full week of evals. Flexibility here doesn't come at the price of a fair rate.
Mindful AI, by design
We share the public's unease with reckless AI deployment. Careful, intentional training data is our answer — and the people producing it are treated as the experts they are.
Why we exist
We've sat in your seat
Orion's founders spent years as annotators on the big platforms. We know the frustrations firsthand: rates that change without notice, queues that vanish overnight, feedback that never comes, and no path that leads anywhere.
So we built the company we wished we'd worked for. One where the people teaching AI are treated as skilled professionals on the cutting edge of an exciting field — with the pay, growth, and respect that implies.
We intend to lead this industry toward a better way.
Where everyone starts. Comparison and rubric work across the projects you're qualified for, with calibration feedback that helps you grow instead of grading you in the dark.
Audit submissions, calibrate guidelines, and mentor taskers — at a higher rate. Reached through the paid Reviewer Readiness Program, not an opaque queue.
Own a project end to end: guidelines, QA strategy, client calibration. Our leads sit in the meetings that shape how models get trained.
What the work looks like
Response comparisons
Read a prompt and two AI responses side by side. Choose the better one and explain why in a sentence or two. Typical task: 2–5 minutes.
Rubric evaluations
Score a single AI response against criteria like accuracy, clarity, and safety, with optional written feedback. Typical task: 3–7 minutes.
And as you grow: reviewing others' work, calibrating guidelines, and paid trainings toward specialist domains — code, medicine, law, safety — that unlock higher-rate projects.
Apply to join
Tell us about yourself — expertise, annotation experience, and what you're looking for. We review every application personally, and when there's a fit we set up your account and onboard you as part of the team. No self-serve queues, no black holes.