From sign-up to your first task

Six steps from a new account to project work, and whether each one is open in the current preview.

Access is decided on the server at every step, not by what the page shows.

The steps

Training opens with your account. Practice tasks and the qualification open once training is done.

Readiness for paid projectsExample account

Practice tasks need only the training.

Create an account
Register with your name, email address and a password. Your dashboard opens straight away, and the application, training and practice tasks all live inside it.
Open now
Fill in the application
Five sections: profile, education, skills, availability and a final review. Each one saves as a draft. An administrator reviews submitted applications and records a decision; approval is never automatic, and you can withdraw at any time.
Open now
Learn the four kinds of work
Training runs in order: text classification, AI response evaluation, pairwise comparison and image annotation. Each module mixes short lessons with practice questions that are scored as you go.
Open now
Practise on sample tasks
Completing training opens the practice marketplace. The tasks use synthetic data and are checked by the same rules a real project would use. Any balance they show is simulated and cannot be paid out.
Open now
Take the General Qualification
Ten questions across all four areas in 30 minutes. You need 80% to pass and have three attempts, with a short wait between them. Image questions pass when your box overlaps the reference by at least half.
Open after training
Onboard for paid projects
Paid work would need an approved application, completed training, identity verification, a signed agreement, a verified payout method and the project’s own qualification. Identity verification exists only as a practice version, and no paid projects are live.
Not open yet

The rules that keep work and its credit unambiguous are built in below the interface.

One owner per task: A task is claimed inside a database transaction, so two people cannot hold the same assignment.
One submission: Ownership checks and unique records stop a task being submitted, or credited, twice.
Practice stays practice: Practice tasks are kept apart from project work in the data itself, not only in how they look.
Decisions are kept: Each review records the checks, the reviewer’s decision and the ledger entry that follows from it.

An application is a request to be considered for future projects. The Careers page sets out what it asks for and what it does not promise.