A fellowship sounds like something you get accepted into and then begin working. Handshake AI is more complicated than that.
You can register, complete your profile, and still spend weeks without a project. You might be invited to an assignment and discover there is another assessment to pass. And even after you start, the amount you earn can depend on whether you’re paid for time, for an approved task, or for several stages of review.
That does not make Handshake AI illegitimate. It does mean the advertised hourly rate is only one small part of the story.
This review looks at how the Handshake AI Fellowship actually operates, from eligibility and project matching to task approvals, payment calculations, and the experiences contributors report.
Table of Contents
Handshake AI at a glance
| Feature | What to know |
|---|---|
| Operator | Handshake, the company behind the career networking and recruiting platform |
| Type of work | AI response evaluation, expert prompts, content correction, research and related tasks |
| Contract type | Independent contractor, with project-specific terms |
| Standard eligibility | US-based applicants with supported work authorization; international participation is currently invite-only |
| Advertised pay | Varies widely by role; public listings show ceilings from $17/hour for general evaluation to $150/hour for some engineering roles |
| Pay models | Hourly, per task, hybrid, or progressive per task |
| Payout setup | Stripe, along with required identity and tax information |
| Main uncertainty | Joining the program does not guarantee project matching or paid assignments |
Handshake AI’s public education-partner materials describe typical hourly offers of $30 to $125, depending on expertise and the project. Its broader job listings also display lower and higher advertised ceilings. Neither figure should be treated as a guaranteed rate for every fellow.
What is the Handshake AI Fellowship?
Handshake is widely known as a career platform used by students, graduates, and employers. Handshake AI is a separate program within the company, connecting people with AI labs that need human expertise to improve their models.
The word fellowship can be misleading if you’re expecting a traditional academic appointment, scholarship, or salaried job. Here, participants are generally independent contractors working on projects that can start, pause, or finish according to an AI lab’s needs.
The work often involves writing challenging prompts, evaluating model answers, spotting factual or reasoning errors, and improving responses using subject knowledge. A finance analyst might assess financial reasoning. An engineer might develop a realistic technical problem. A researcher might verify citations and whether an answer genuinely follows a question’s constraints.
Not every project requires previous AI training experience. Some require a bachelor’s degree or general analytical ability; others look for master’s degrees, PhDs, postdoctoral research, or substantial professional expertise. The qualifications attached to a specific opening matter more than a broad description of the program.
Getting accepted is not the same as getting a project
This is the most important point for first-time applicants.
Handshake does not operate every project as a self-service task board where signing up immediately unlocks paid work. Its team matches fellows to opportunities based on their education, expertise, profile, and the current demand of partner AI labs. The company does not promise a fixed waiting period for a project invitation.
The process generally follows this sequence:
- Apply and complete your profile. Register through Handshake, verify your phone number, provide education and work history, upload a resume, and accept the applicable terms. You can skip connecting a school if you’re joining through the open network.
- Wait for a suitable project match. Your profile is considered as opportunities arise. A complete profile improves the information available for matching, but it cannot create demand where none exists.
- Express interest when invited. An invitation may describe the project, expected duration, time commitment, and compensation. Expressing interest is not itself final selection.
- Complete project screening and onboarding. Some projects require a skills assessment, training exercises, work samples, or additional checks. Passing one assessment doesn’t ensure entry to another project.
- Start authorized tasks. Paid work begins under the project’s rules once you’re cleared to participate. Work outside the approved process may not be eligible for payment.
Handshake’s own guidance says failed project assessments often cannot be retaken, although failing one usually doesn’t remove your profile from consideration for unrelated projects.
If you’ve experienced long waits on other platforms, our article on why AI training applications don’t always lead to work discusses the wider problem of screening without guaranteed assignments.
Who is actually eligible to join?
Despite promotional language about remote, flexible work, Handshake AI’s standard program is US-based. Its current eligibility guidance requires eligible US applicants to have a Social Security Number or Individual Taxpayer Identification Number and work authorization that Handshake supports.
US citizens and lawful permanent residents qualify under the published policy. Some F-1 students may participate through approved CPT or OPT arrangements, but they must resolve their authorization with their designated school official or the relevant immigration authority before accepting a project. The published rules say STEM OPT extensions, J-1, H-1B, and TN arrangements are not supported, along with certain other statuses.
For people living outside the United States, the policy states that international fellows are currently invite-only. An international invitation should be verified through the program’s official channels. Applicants should not assume that a referral link overrides these restrictions.
The travel policy adds another limit. US-based fellows generally need prior written approval to perform project work from outside the US, and location-masking tools are prohibited.
Readers in India, Nigeria, Kenya, the Philippines, or elsewhere can use our country eligibility guide for AI training jobs to identify other platforms worth researching. It would be misleading to present Handshake’s standard application as universally open worldwide.
Four ways Handshake AI pays for work
One of the biggest changes in the program is that different projects now follow different compensation structures. Handshake’s payment guidance, updated November 6, 2026, describes four models.
| Model | How earnings are calculated | What to watch |
|---|---|---|
| Hourly | Pay for logged time on submitted tasks, subject to task and possibly weekly caps | Extra time beyond the applicable cap may not be paid |
| Pay per task | A fixed amount is earned when a task reaches Ready to Deliver (RTD) | Submission alone does not trigger payment |
| Hybrid | Hourly pay plus an additional amount when the task reaches RTD | Both time caps and review approval can matter |
| Progressive pay per task | Payments accrue at defined stages, from submission through review and RTD | The largest final portion may depend on completing every review stage |
RTD, or Ready to Deliver, means the task has cleared the required review process and is approved to go to the client. Review and revision delays can therefore affect when task-based earnings appear.
Here’s why the structure matters. Imagine a project offers $150 per accepted task. If you spend three hours writing and revising a successful submission, that’s $50 per hour before expenses and taxes. If the task takes ten hours, the effective rate becomes $15 per hour. If it never reaches the required approval status under a standard pay-per-task agreement, you may not receive that task payment at all.
These are illustrative calculations, not an example of a particular Handshake contract. The lesson is to evaluate the approval conditions and realistic time commitment, not just the amount shown on the project card.
The same issue applies to hourly work. Handshake says hourly projects can cap payable time per task. You might genuinely spend longer on a difficult submission without being paid for every additional minute. Our article on what AI training jobs actually pay explains why effective earnings often differ from headline rates.
Is onboarding or assessment time paid?
Not automatically. This deserves attention before you agree to lengthy training.
Handshake’s general earning policy says some project onboarding is compensated and some is not. Payment must be stated in the specific Project Terms. Basic account setup, such as providing payout information or completing initial administrative checks, is not payable project time. Optional office hours and general guidance sessions are also normally unpaid.
Some of Handshake’s education-partner materials describe compensated onboarding for selected fellows, but that general description should not override the agreement for your particular project. Before starting any assessment or training exercise, check the exact terms and whether completing it creates a payment entitlement.
There is also an AI Skills Studio for practice and skill development. Handshake states that these missions are not paid work. Preparation may be useful, but it’s not the same as earning from an assigned contract.
Payment schedule, Stripe, and the new earnings dashboard
Handshake AI uses Stripe for fellow payouts. Contributors need to complete their identity verification, tax details, and payout setup before payments can be processed.
For projects on the standard weekly schedule, Handshake says it generally sends payouts by Wednesday afternoon for the previous Monday-through-Sunday work period, with processing to the payout provider often taking another 48 hours. Some projects have different terms, particularly when payment depends on a task reaching a review milestone.
A dashboard estimate isn’t the same as cleared earnings. Handshake reviews eligible work and may adjust logged hours for task time caps, weekly limits, inconsistent activity records, invalid submissions, or approved disputes.
An updated Payments dashboard began rolling out on October 14, 2026, and Handshake refreshed its explanation on November 6. It is intended to show pay by week and project, including base pay, bonuses, task IDs, adjustments, and the reasons for deductions. The rollout is phased, so not every fellow necessarily sees the new view yet.
The company also describes a dispute process for missing or incorrect pay. If the payment looks wrong, check the project’s terms, the task status, and the relevant pay week, then use the official support or payment dispute channel. Keep your own permitted record of task IDs and approvals, but don’t copy confidential client materials.
What contributors report about the experience
Handshake AI has real company infrastructure, documented contractor policies, and an established payout process. That answers part of the legitimacy question. It does not mean every participant has a smooth experience or steady work.
Public reviews and Reddit discussions contain accounts of contributors who received high-paying assignments and payments, alongside complaints about slow project matching, abrupt removals, disputed task approval, and support delays. Some people say a seemingly attractive per-task amount became far less attractive after many hours of revisions. Others describe waiting for available tasks even after onboarding.
These are individual, often anonymous accounts that cannot be independently verified. Reviews on general Handshake profiles can also mix the original career platform with the AI Fellowship, so headline review scores are a poor substitute for studying project-specific experiences.
There is one useful point where the company’s documentation and contributor frustrations meet: Handshake itself warns that projects can pause or end with little notice, and that continued participation depends on project-specific quality standards. It also acknowledges that a pay estimate may be adjusted after verification.
Those facts don’t prove every complaint. They do establish that uncertain task supply, review requirements, and payment adjustments are risks prospective fellows should understand before committing substantial time.
What I would check before accepting an invitation
A strong application is one part of the process. A sensible contract decision is another.
Before accepting project work, ask yourself whether the pay is hourly or conditional on review, whether training is compensated, what the time caps are, and how long tasks usually remain in review. Check the project duration, the expected weekly commitment, and whether task supply is guaranteed. Look closely at what happens if a project ends while your submissions are being evaluated.
You should also understand the rules on external AI tools. Handshake’s education-partner guidance prohibits using external AI assistants such as ChatGPT or Claude to generate or edit project submissions unless the specific authorized workflow says otherwise. Violating project policies can affect both access and payment.
If your financial plans depend on predictable weekly income, don’t count potential project hours as income until you’ve actually been assigned eligible work. Even a high hourly offer is not helpful during weeks with no tasks.
Is Handshake AI worth joining?
Handshake AI is a legitimate program worth considering for eligible US-based students, graduates, and specialists who can tolerate uncertain project availability. It can offer intellectually interesting AI evaluation work and potentially attractive compensation for difficult subject-matter tasks.
But there are meaningful limits. The standard program is not broadly open to international applicants, an accepted profile isn’t a project, and the payment model can materially change your effective earnings. Anyone considering a pay-per-task assignment should read the approval milestones particularly carefully.
If you meet the current eligibility rules, you can explore the Handshake AI Fellowship. If you don’t, it makes more sense to focus on opportunities that support your location and expertise rather than attempting to work around the restrictions.
For comparison, Mercor offers a different expert-contracting route, with its own screening and geographic requirements. Our comparison of AI training platforms can help you research other options.
The useful question isn’t simply whether Handshake AI pays well. It’s whether the specific project you’re offered pays fairly for the work, review burden, and uncertainty you’re accepting.
Research updated November 9, 2026. Handshake’s project eligibility, published pay structures, and payment policies may change. This review is based on company documentation and attributed public contributor reports, not firsthand employment with Handshake AI.
Disclosure: This article contains referral links. I may earn a commission when eligible applicants complete qualifying activities, at no additional cost to them. Referral links do not guarantee selection, paid assignments, or earnings.
Leave a Reply