Comparisons

Is Handshake AI Legit? Pay, Eligibility and Reviews

Is Handshake AI legit? Yes. Its AI Fellowship pays US students, grads and pros $17 to $150 an hour to train AI. Who qualifies, how pay works, and the catches.

Comparisons

On this page

Yes, Handshake AI is legit. It's the AI training program run by Handshake, the career network many US college students already use for internships and jobs. Through the Handshake AI Fellowship, students, recent graduates and professionals get paid to review and improve AI answers for AI labs. That doesn't mean it's steady work. It's US-only, projects start and stop on short notice, and reviews about organization and pay disputes are mixed. Here is who qualifies, what it pays, and what to expect.

Disclosure: Handshake AI isn't one of our referral partners. Some micro1 and Mercor roles on our AI training jobs page use referral links, and we may be paid if you are hired through them.

TL;DR

Handshake AI is a real, paid AI training program from an established company, not a scam. You must be based in the US with valid work authorization and at least 18. Listed roles run from about $17 an hour for general AI evaluation up to $150 an hour for software engineering, with most specialist roles between $80 and $125. Pay goes out weekly, usually by Wednesday, through Stripe. There are no guaranteed hours, and projects can pause based on what the AI labs need. If you're outside the US, or want more options, browse micro1 and Mercor roles on our directory.

What Handshake AI is

Handshake started as a career platform connecting college students with employers. More recently it launched Handshake AI, which uses that network to supply AI labs with people who can judge AI answers in their field. Handshake's help center describes the program as connecting "current students, recent graduates, and professionals" with flexible, project-based work to help train AI models.

The work is what you'd expect from AI training:

  • Reviewing and editing AI-generated answers
  • Annotating datasets
  • Writing better answers and suggesting how a model's output should improve

The AI projects are separate from the career platform. Handshake says they don't involve screening job candidates, and the AI labs you work for aren't named.

The program is large. Handshake's own site says it has more than 188,000 fellows and has paid out over $500 million. Third-party estimates from Sacra put Handshake's yearly gross revenue at around $1 billion in 2026, mostly from AI training. Those figures are estimates, not audited numbers, but they show this is a major business, not a side project.

Who can apply

This is where Handshake AI differs most from competitors:

  • US only. You must be based in the United States, with work authorization Handshake can support. Some F-1 students eligible for CPT or OPT may qualify. STEM OPT isn't supported.
  • 18 or older.
  • Students, graduates and professionals. Roles list requirements from an associate's degree up to a PhD or postdoc, depending on the project. No prior AI experience is needed.

If you live outside the US, Handshake AI isn't an option. Mercor and micro1 both hire internationally for many roles, and our AI training jobs page hides roles that aren't open in your country.

What Handshake AI pays

Handshake's homepage advertises "up to $100/hr," but the roles it lists show a wider range:

Example role on Handshake's siteListed pay
AI evaluation specialistUp to $17 an hour
Energy professionalUp to $80 an hour
FP&A analystUp to $80 an hour
Philosophy expertUp to $120 an hour
Game developerUp to $125 an hour
Software engineerUp to $150 an hour

You see a project's rate and any bonus before you join it. Projects pay in a few ways: by the hour (capped per task), by the task once it's approved, or a mix of both. Handshake may also set weekly hour caps.

How you get paid: the pay week runs Monday to Sunday (Pacific time), and Handshake typically sends payouts by Wednesday afternoon. It can take about 48 hours to show up. You set up payouts through Stripe and fill out a tax form. You're an independent contractor, so you handle your own taxes. We found no fees charged to fellows.

How getting started works

  1. Apply and verify your identity.
  2. Get matched to a project based on your background and what the labs need.
  3. Take the project training, and pass an assessment if the project requires one.
  4. Do the work on your own schedule.
  5. Get paid weekly.

Projects open on a rolling basis, so how fast you start depends on demand in your field.

The complaints, and how seriously to take them

Reviews on Glassdoor and Indeed range from one star to five, and the patterns are consistent:

  • Uneven work. Handshake says projects "may start, pause, or end" based on the labs' needs, "sometimes with short notice." Some months are busy, others are empty.
  • Disorganization. Several reviewers praise the work and timely pay but describe changing instructions and slow support.
  • Pay disputes. Some fellows report pay that was wrong or late and had to file a dispute. Others say disputes usually get resolved. Handshake has a form for missing payments.
  • Pay terms can change. Handshake says a project's pay terms can change over time, with notice in advance.
  • Removal for inactivity or quality. Inactive fellows may lose project access, and fellows who don't meet expectations can be removed.

None of that makes Handshake AI a scam. It's flexible contract work: good pay when there's a project, nothing when there isn't.

Who Handshake AI is a good fit for

A good fit: US college students, grad students and recent graduates, especially in STEM, finance, philosophy and other specialist fields. It's also good for US professionals who want a few flexible hours a week.

A weak fit: anyone outside the US, anyone on STEM OPT, and anyone who needs dependable weekly income. If you're early in your career and new to AI tools, learning to use them well is worth more than any single gig. Our beginner's guide to using Claude and free course cover that.

How to stay safe

  • Never pay to join. Handshake AI doesn't charge fellows. Anyone asking for a fee is impersonating it.
  • Apply through Handshake's own site. Fake recruiters copy the names of AI training programs.
  • Share tax and ID details only during official onboarding.
  • Apply to more than one platform. Work on any one is uneven. See our Mercor review, micro1 review and the best AI training jobs compared.

Frequently asked questions

Is Handshake AI legit or a scam?

Handshake AI is legit. It's the AI training program of Handshake, an established career network for students, and it pays fellows to review and improve AI answers for AI labs. The real downsides are uneven project work and occasional pay disputes, not a scam. Handshake AI doesn't charge fellows any fee.

How much does Handshake AI pay?

Handshake AI roles listed on its site range from about $17 an hour for general AI evaluation to $150 an hour for software engineering. Many specialist roles list $80 to $125 an hour. You see each project's rate before you join.

Can you do Handshake AI outside the US?

No. The Handshake AI Fellowship requires you to be based in the United States with valid work authorization. Some F-1 students eligible for CPT or OPT may qualify, but STEM OPT isn't supported.

How does Handshake AI pay you?

Handshake AI pays weekly. The pay week runs Monday to Sunday, and payouts are typically sent by Wednesday afternoon through Stripe, then can take about 48 hours to arrive. Fellows are independent contractors and handle their own taxes.

Do you need experience for Handshake AI?

No AI experience is needed. Roles do list education requirements, from an associate's degree up to a PhD or postdoc, depending on the project. Some projects also require passing an assessment.

Find more AI training work

Handshake AI is a strong option if you're a US student or graduate. To compare other paid AI training roles, including ones open outside the US, browse our AI training jobs directory, or use the AI job matcher to find the best fits for your background.

← All guides


MIT 2026 © Nextra.