Handshake AI assessments and onboarding: what to expect, retake rules and identity checks

Why this matters

On Handshake AI, the screening that decides whether you work on a project happens after you are invited, not when you apply. Handshake's help centre describes a sign-up form, a wait to be matched, and then, project by project, an onboarding stage that can include training modules, a quiz, a reviewed practice task and an identity check. Its assessments article says that if you fail, "in most cases, retakes are not available" for that project.

This guide covers what Handshake publishes about assessments, onboarding and identity checks, with worker reports kept separate. It does not repeat eligibility, matching odds or pay rates; those are in our guide to who qualifies for the Handshake AI fellowship and what it pays. For the platform overview, see the Handshake AI company page.

One thing to say up front: none of the Handshake AI help-centre articles we read describes an interview, AI or human, as part of the process. What they describe are assessments, training and, for some projects, portfolio or work-sample reviews.

The application flow, in order

Handshake's "Apply to the Handshake AI Fellowship" and "From application to first project" articles describe these steps:

1. Sign up. Log in or create a Handshake account, optionally connect a school, and verify your phone number with a one-time code.
2. Answer the work authorization questions: whether you are authorized to work in the US, and whether you need CPT, OPT or H-1B sponsorship.
3. Complete your expertise and profile. Your résumé must be a PDF of no more than 50 MB. Then accept the Terms of Service, Privacy Policy and Contractor Agreement.
4. Wait to be matched. The apply article says "there's nothing else you need to do while you wait to be matched", and the journey article says "there is no fixed timeline for receiving a project invitation."
5. Get an invitation and express interest. The email describes the project, its length, time commitment and compensation. Some projects also ask for "a slide deck, portfolio, or other work samples."
6. Selection. After your interest form, Handshake describes "a short waiting period while interested candidates are reviewed", then an acceptance email if you are chosen.
7. Project onboarding. Required steps are listed in the project workspace and vary by project. This is where assessments usually sit.
8. Training, then tasking. Training covers project details, guidelines and time tracking. You start tasking once approved.

Identity verification and payout setup are both required before project work, according to the apply article.

The "Next steps after applying" article adds that "some projects may require additional evaluations, training exercises, or portfolio reviews before selection", so a project can screen you before choosing you and again during onboarding.

What the assessments actually are

Handshake's "Assessments on Handshake AI" article (updated 27 September 2026) is the main source here. It says assessments are "project-specific, and the format, length, and passing criteria will vary depending on the partner lab and the type of work involved." Details arrive with your project onboarding materials.

Before tasking, it says, you may be asked to complete one or more of:

• "A knowledge quiz based on your onboarding training materials"
• "A practice task or sample submission reviewed by the Handshake AI team"
• "A written or skills-based evaluation relevant to your project domain"

The same article says assessments "are designed to reflect the real work you'll be doing on the project, so your onboarding materials are the best preparation."

The general onboarding flow

An older Handshake article, the fellowship overview written for university partners (last updated 16 January 2026), gives a more concrete picture. It says every accepted fellow "goes through our general onboarding flow, which has ~2 hours of learning modules on Canvas. They then complete a knowledge quiz, and are passed along to a live practice task where they get feedback from reviewers." It adds that each project sets its own onboarding and training on top of that.

Because that article is older than the rest, treat the Canvas detail as Handshake's description at the start of 2026 rather than a guarantee of what you will see now.

Assessments do not stop at onboarding

The assessments article says assessments "may also appear during a project if quality standards need to be re-verified." Handshake's "Fellowship expectations" article goes further: "New assessments may be introduced as project needs evolve, and continued participation may depend on meeting them." Passing once does not settle it for the life of a project.

What happens after you submit

Handshake says there may be "a processing period" while the Handshake AI team or project reviewers evaluate your submission, that this is normal, and that "you do not need to resubmit." You are notified when a decision is made. It does not say how long the processing period lasts.

Retake and attempt rules

From the assessments article:

• A fail ends that project's onboarding. Failing means you "did not meet the minimum score or quality threshold." If retakes are not permitted, "you will not be able to continue onboarding for that specific project."
• Retakes are rare. "In most cases, retakes are not available." Where they exist, availability "is determined on a per-project basis and is not guaranteed."
• Wanting a better score is not a reason. Handshake says that "simply wanting to improve your score is not typically grounds for a retake" and that requests are "generally only considered when there is a documented error or technical issue."
• A fail is not the end of the fellowship. Handshake says failing "does not disqualify you from the Handshake AI Fellowship altogether. Your profile remains active and you'll still be considered for other projects that match your qualifications."

If something breaks mid-assessment (it gives the assessment not loading, a submission error and being locked out as examples), Handshake says to contact Support and "do not attempt to restart or resubmit on your own, as this may affect your results." Since a documented technical issue is the main ground for a retake, take a screenshot of the error before you write in.

Rules that apply during assessments and onboarding

Handshake's "Original Work and Prohibition on Usage of LLMs Policy" covers all work product, and it is stricter than many platforms' rules. It prohibits using large language models or other AI-assisted tools for developing prompt ideas, generating or editing content, solving tasks, and even "using LLM-based grammar or writing tools for stylistic improvements, grammar correction, or formatting", unless Handshake authorises a tool in writing. It also says prompts taken from Khan Academy, Chegg, textbooks or "your annotation work on other platforms" do not count as original.

The policy asks you to keep preparatory materials (drafts, notes, ideas) for at least 90 days after delivery, because Handshake may ask to see them. If Handshake believes an LLM was used, payment for that work is suspended pending review, and you may be given a chance to show you did not use one. If it concludes you did, the work is rejected and unpaid, and you may be removed from the project or the programme.

Handshake's "Project Removal" article says that if you are removed from a project, you can request an appeal and support will follow up by email.

Is onboarding paid?

Handshake's articles disagree, so go by the newest one and your Project Terms.

• The January 2026 university-partner overview says fellows "are compensated for project onboarding time", and that the general Canvas, quiz and practice-task flow is compensated too.
• The "Earning on Handshake AI" article (updated 13 September 2026) says "some projects include paid onboarding time, while others do not", and that "if onboarding pay is not explicitly mentioned, time spent onboarding is not paid." It also says the initial onboarding checklist (payment details, project access requirements and account setup) is "not considered payable project time and should not be tracked."

The Earning article tells fellows to read the Project Terms before starting onboarding, and warns that logging ineligible time can lead to non-payment or removal from the project.

Identity verification

Identity checks run through Persona. Handshake's "Identity Verification: How It Works" article (updated 1 October 2026) calls it "a one-time requirement that must be completed before you begin your first project." You can sometimes complete it as soon as you finish your application, but Handshake says this "does not increase your chances of being matched." Some projects add a background check; see the fellowship guide for that.

What Handshake publishes about the check itself:

• Accepted IDs: driver's license, state ID, passport or Permanent Resident Card.
• Device: a smartphone or tablet with a working camera. The troubleshooting article says "computer-only verification attempts may fail." You start on the computer and scan a QR code to continue on the phone.
• Live capture only. Photos of the front and back of your ID, then a selfie. Handshake's "Getting Verified and Staying in Good Standing" article says not to submit "screenshots, edited images, photocopies, or another person's identification."
• Browser: if Persona will not load, the troubleshooting article suggests the latest Chrome or Safari, disabling ad blockers or privacy extensions, and moving from a work or school network to personal Wi-Fi or a mobile hotspot.

The good-standing article (updated 28 September 2026) lists what keeps your access from being paused:

• One legal name everywhere. Your full legal name must match across your Handshake AI profile, your Handshake account, your ID and your Persona submission. No nicknames, shortened names or alternate spellings.
• No location masking. Do not use a VPN, proxy, remote desktop or location-masking browser extension when you access Handshake AI or do the work.
• One account. Do not open a second account to fix a profile problem, redo verification or change location; contact support instead.

The "Identity verification and account authenticity" policy adds that Handshake may request "periodic re-verification", and that violations can lead to restricted access, withheld payment, removal from projects, suspension or permanent deactivation. Our identity verification guide covers why platforms check these things.

Practice options

Handshake publishes no practice assessment and no sample questions that we could find. The closest official advice is in "Next steps after applying", which suggests that while you wait you build experience in your field, read up on AI model evaluation, reasoning and LLM workflows (it points to an explainer on reinforcement learning from human feedback), and build projects with Handshake's AI Showcase tool. Once you are on a project, the onboarding materials are the preparation Handshake recommends.

What workers report

This section is what workers say, not what Handshake publishes. It comes from Handshake's Trustpilot page, read on 1 October 2026 with a keyword filter, so treat it as anecdote rather than a measure of how common anything is.

• Unpaid assessments and training. A one-star reviewer in May 2026 complained that Handshake does not pay for assessments. A two-star reviewer in July 2026 described spending more than two hours on unpaid training. Both fit the Earning article's rule that onboarding is unpaid unless the Project Terms say otherwise.
• Many assessments, unexplained fails. A one-star reviewer in June 2026 described a stream of assessments and being told they had failed them.
• AI-use flags with little detail. A two-star reviewer in September 2026 said they were removed from a project after submissions were flagged for possible LLM use, asked for documentation of their work process on appeal, and never shown which submission was flagged.

Prep checklist

Each item is tied to something Handshake publishes:

• Make your profile complete and specific. Matching is "based on project needs, qualifications, experience, and area of expertise", so your profile is what gets you invited. Our AI training résumé guide helps
• Have a portfolio, slide deck or work samples ready, since some invitations ask for them
• Read the Project Terms before onboarding to see whether onboarding is paid
• Do every training module before the assessment. Handshake calls the onboarding materials "the best preparation" and says quizzes are based on them
• Set aside uninterrupted time. Handshake's own tips say not to rush and to follow each project's instructions carefully
• Treat the first attempt as the only one. Retakes are mostly unavailable, and a better score is not grounds for one
• If it breaks, stop. Screenshot the error and contact Support rather than restarting or resubmitting
• Do not use any AI tool, including AI grammar checkers, unless Handshake has authorised it in writing, and do not reuse prompts from textbooks, Chegg, Khan Academy or other platforms
• Keep your drafts and notes for at least 90 days
• Prepare for Persona: a phone or tablet with a camera, a valid unexpired ID, good light, no VPN, and the same legal name everywhere
• Expect more assessments later on a project, and keep up with guideline changes

For general technique, see how to pass AI training qualification tests. If you pass and still get no work, accepted but no tasks explains why that happens.

What Handshake AI does not publish

• No pass marks or rubrics. Criteria vary by project and are not published
• No assessment lengths, beyond the January 2026 description of about two hours of general Canvas modules
• No timeline for results. The assessments article mentions a processing period without a duration
• No pass rates or acceptance rates
• No feedback policy for failed assessments
• No practice assessments or sample questions
• No interview step in any help-centre article we read. If someone claims to be scheduling a Handshake AI interview, check the message against Handshake's list of genuine channels, covered in the fellowship guide

Sources

• Handshake AI help centre: Assessments on Handshake AI
• Handshake AI help centre: Apply to the Handshake AI Fellowship; From application to first project; Next steps after applying
• Handshake AI help centre: Handshake AI Fellowship Overview and FAQs (for EDU Partners), for the Canvas modules, quiz and practice task
• Handshake AI help centre: Handshake AI Fellowship expectations; Earning on Handshake AI; Original Work and Prohibition on Usage of LLMs Policy; Project Removal
• Handshake AI help centre: Identity Verification: How It Works; Troubleshooting Persona identity verification; Getting Verified and Staying in Good Standing; Identity verification and account authenticity
• Trustpilot (joinhandshake.com), for worker reports, attributed above

Last checked 1 October 2026. Worker reports are attributed to Trustpilot reviewers and are not Handshake's statements. We are not affiliated with Handshake and we do not process applications.