Handshake

For a decade, Handshake was the thing you signed into freshman year to find an internship. Garrett Lord built it the slow way, driving a Ford Focus campus to campus to sign up universities one school at a time, and by 2024 it connected 25 million students, 1,600 schools, and a million employers.

Then the story flipped. In early 2025 the company launched Handshake AI, put its network of PhDs and specialists to work labeling data for frontier AI labs, and rode that to roughly a $1B gross run rate inside a year. The career network for the AI economy now makes most of its money teaching the AI.

Handshake runs a three-sided marketplace: job seekers on one side, employers on another, and the college career centers that vouch for the students in between. Dynamic profiles, job matching, and skill-building tie the three together, and the company's pitch is that talent should not be gated by who you know or where you went to school.

The newer half is Handshake AI, which turns that same pool of educated members into a contractor workforce that creates evaluations, reasoning chains, and domain-specific training data for the labs building frontier models. Same members, very different product.

The original problem was a recruiting system that favored a handful of elite schools and left everyone else guessing. Handshake fixed the sourcing by pooling students, employers, and career centers in one place so a company could reach talent it would never have visited in person.

The AI side solves a different sourcing problem. Public and synthetic data have run out of headroom, so labs now need proprietary human judgment in narrow domains like analog circuit design, cartography, or audio engineering. Handshake already had those experts on the platform, so it cut out the middleman and sold their expertise directly.

The consumer side is built for early-career job seekers, the students and recent grads who do not have a network yet, plus the career-center staff and the million-plus employers hiring them. That is the base the whole company grew from.

Handshake AI aims much higher up the value chain. Its buyers are the frontier AI labs and enterprises that need proprietary training data, and its suppliers are the 30,000-plus specialists and PhDs who get paid to produce it. The AI Trainer gigs, meanwhile, court experienced professionals who want flexible, remote side income rather than a full-time job.

Handshake ships two very different things under one roof. The consumer platform is the career network people know, with matching, profiles, communities, and events. Handshake AI is the fast-growing data business, split between selling trained human judgment to labs, deploying engineers inside enterprise accounts, and running the AI Trainer marketplace that supplies the labor.

The common thread is the network. Every product leans on the same 25 million members, which is the moat none of the pure-play data vendors started with.

The core marketplace is free for students and career centers and monetized through employers, the classic recruiting-network model, and it still throws off roughly $190M a year. The real growth engine is Handshake AI, which sells human-generated training data and evals to frontier labs on enterprise contracts.

That second business is high volume but thin margin: Handshake pays contractors around 70% of gross revenue, so a ~$1B gross run rate nets closer to $300M. Pricing is contract-based and not published; there is no public pricing page.

Handshake walked into a fight already in progress. Scale AI and Surge AI both crossed $1B in gross revenue selling AI training data, and Mercor, a three-year-old rival, hit the same mark in early 2026. Handshake's twist is that it was the source those competitors recruited from before it went direct.

Its edge is the network it spent a decade building: 25 million vetted members and 1,600 school relationships that a pure-play data vendor cannot replicate overnight. The risk is that data spend is the bet, and if labs pull back or margins compress, a business paying out 70% of revenue has little cushion.

The scale is real and easy to check. The network reaches 25 million knowledge workers, 1,600-plus educational institutions, and a million employers, including every Fortune 500.

The AI business is the eye-catching part: from a standing start in January 2025 to roughly $1B gross annualized revenue by April 2026, paying out about $60M a month to more than 30,000 individuals. Handshake says it tripled its ARR at scale in 2025.

The AI Trainer gigs draw praise for what you would expect: flexible remote hours, legitimate projects, and pay that runs from roughly $25 an hour up to $125 for specialized work. For the right expert it is real side income on your own schedule.

The recurring gripe is inconsistency. Contributors report long gaps between assignments and mixed communication, so it reads as supplemental income rather than a dependable full-time wage. Entry is also selective, tilted toward degreed specialists over beginners.

Handshake has raised about $434M across six rounds, capped by a $200M Series F in January 2022 that set a $3.5B valuation, led by Coatue Management and Valiant Peregrine Fund. Earlier backers include Kleiner Perkins, Spark Capital, EQT Ventures, GGV Capital, Base10 Partners, the Chan Zuckerberg Initiative, and Omidyar Network.

The interesting part is that the funding history now understates the company. The AI business alone is running near $1B gross, which is a different scale than the valuation the last round implied.

Handshake was founded in 2014 by Garrett Lord, Ben Christensen, and Scott Ringwelski, with Lord still CEO and setting a deliberately intense tone. The values say it plainly: students first, Olympic pace, own the outcome, leave nothing to chance.

The AI side reads like a startup inside the company, staffed with people from Scale AI, Meta, xAI, Palantir, Notion, and Coinbase, and the engineering roles are explicit about five days a week in the San Francisco office. Perks are strong and equity comes with most full-time roles, but this is a company that wants people who move fast and finish.

Full-time roles cluster in the six figures. Disclosed engineering bands run roughly $150K to $325K, with sales, product, marketing, and G&A landing in the low-to-mid six figures depending on level and location, all in USD with a small number of INR bands for the Bangalore office.

The AI Trainer contractor work is hourly instead, spanning roughly $25 to $125 an hour by specialty. Equity comes with most full-time positions, and sales roles carry variable upside on top of base.

The legal entity behind the brand is Stryder Corp, doing business as Handshake, which the privacy policy names directly. Because the platform handles student records shared by schools, it processes that data in compliance with FERPA, and it references GDPR and CCPA obligations in its data-partnership and contractor terms.

Contractor onboarding is heavier than most: know-your-customer identity checks, government ID and selfie verification, and tax and work-authorization collection through third-party providers, reflecting the money it moves to a large 1099 workforce.

The coverage in 2025 and 2026 is almost entirely about one thing: how a college job board became a billion-dollar training-data business seemingly overnight. Trade press and analysts frame Handshake alongside Scale, Surge, and Mercor as the companies now selling AI its expertise.

The links below trace both chapters, from the 2022 Series F that made it Gen Z's LinkedIn to the 2026 reports on its ARR crossing $1B.