AI training jobs for beginners: which platforms actually take them

The beginner pool, by platform

Of the 699 roles live on this site on 29 September 2026, we tag 74 as entry level and 127 as junior. The rest ask for professional experience, often a lot of it. If you are looking for AI training jobs for beginners, those 201 listings are the realistic pool, and three platforms post all of them:

Alignerr has only senior roles on the board right now, and we list no open roles from Outlier or Handshake AI, so we cannot say what either takes from beginners today.

micro1: the widest entry tier, and the lowest pay

micro1 posts the most entry-level work and the widest spread of it. At the bottom are audio and video annotation roles at $6–10 an hour, such as its Data Annotator, which asks for no degree, no years of experience and no AI background. Above that sit transcription and language roles, mostly $10–36, bilingual expert roles at $30–65, and a few generalist roles such as the Evaluation Specialist for recent graduates at $20–60, which wants precise written English but says it need not be your first language.

A large share of micro1's entry tier is physical data collection: filming household tasks, recording video at home, working alongside robots. Most of those are open to US residents only.

Mercor: better pay, rarely unqualified

32 of Mercor's 35 entry roles start at $30 an hour or more, but entry there usually means new to AI work, not new to working life. Its bilingual AI safety generalist roles ask for a bachelor's degree, completed or in progress. Its Generalist (Must own MacBook) wants a current or recent master's student, or two to three years' experience plus a bachelor's. The roles that skip credentials, such as its ChatGPT user and voice-cloning projects, are mostly tied to one country each.

Invisible: flat rates, open worldwide

All 24 of Invisible's entry and junior roles pay a flat rate and none names a country restriction: $17 an hour for voice, audio and image evaluation, $21 for coding-and-JSON evaluation, and $1.30 a unit on two speaker projects. Its Image Data Contributor and Evaluator states that no professional background is required.

What the beginner pool actually looks like

• Most of it is language work. 119 of the 201 require a language other than English. If you have one, the bilingual guide shows where it pays best; the language jobs page lists them.
• The bottom pays little. 51 of these listings top out at $20 an hour or less.
• Residence matters more at entry level. 17 of the 74 entry roles are US-only, against 2 of the 127 junior roles.
• Junior is a loose label. A few junior listings are professional roles, such as a junior transactional attorney at $70–90. Read the requirements, not the tag.

For the order to apply in, see how to get started, and for country rules, who can actually do this work. The general jobs page collects the generalist roles.

How to apply with no experience

No AI background is not the same as nothing to show. Customer service, editing, content moderation and quality assurance all train the habits these tasks run on: reading a long set of rules, applying them the same way every time, and flagging the item that fits none of them. Name those habits in the application. A line about checking replies against a written policy tells a reviewer more than an interest in AI does. If you speak a second language, say which one and how well; it is a requirement on 119 of the 201 beginner listings.

When a form asks why you are a fit, answer by describing the task. For micro1's Data Annotator, that means labelling audio and video exactly as the guidelines say, including the clips that are hard to make out. For Invisible's image evaluation work, it means judging each image against the brief rather than your own taste. For micro1's Evaluation Specialist, it means precise written English and reasons you can explain.

Before you commit to per-unit work, do the arithmetic. Divide what a task pays by the minutes it really takes, counting the time spent rereading instructions and redoing rejected items, then multiply by 60. That is your effective hourly rate. Invisible's speaker projects pay $1.30 a unit, so to match the $17 an hour on its hourly roles, each unit has to take you about four and a half minutes or less. At ten minutes a unit, the same work pays $7.80 an hour, inside micro1's lowest annotation band. Time your first batch, run the sum, and compare it with the hourly listings above before you decide where your hours go.

Sources

Counts and rates come from the 699 listings on this site on 29 September 2026, each quoting its platform's published rate. Experience levels are our own tags, based on each listing's requirements. Listings change daily and the job page always outranks this guide. We are not affiliated with these platforms and we do not process applications.