Handshake, the AI data-labeling startup best known for its origins as a college recruiting platform, has acquired data auditing company Cleanlab in a talent-focused deal aimed at strengthening its research and data quality capabilities, the companies confirmed.

The transaction primarily functions as an acqui-hire, bringing nine key Cleanlab employees into Handshake’s research organization, including co-founders Curtis Northcutt, Jonas Mueller, and Anish Athalye — all MIT-trained computer scientists. Financial terms were not disclosed.

Founded in 2021, Cleanlab developed software designed to automatically detect labeling errors and inconsistencies in training datasets without requiring additional rounds of human review. Its algorithms help identify mislabeled or low-quality data at scale, a capability increasingly viewed as critical as AI models grow more complex and sensitive to data quality. The startup had raised a total of $30 million from Menlo Ventures, TQ Ventures, Bain Capital Ventures, and Databricks Ventures, and at its peak employed more than 30 people.

For Handshake, the acquisition signals a strategic move beyond pure human-powered labeling services toward a more defensible, technology-driven data infrastructure model.

Originally founded in 2013 as a career platform connecting college students with employers, Handshake built one of the largest early-career talent networks in the United States. Over the past year, however, the company expanded into AI data labeling, leveraging its access to specialized professionals — including doctors, lawyers, and scientists — to provide high-quality human annotations for foundation model developers.

Cleanlab’s research expertise now adds an algorithmic layer to that human network.

“We have an in-house research team that constantly evaluates where our models are weak and how we can improve the quality of our data,” Sahil Bhaiwala, Handshake’s chief strategy and innovation officer, told TechCrunch. “The Cleanlab team has been focused on this exact problem for years.”

The deal also reflects a broader shift underway across the AI training ecosystem. As foundation models mature, the competitive focus is moving away from sheer labeling volume toward precision, consistency, and trustworthiness of data. Companies are increasingly investing in automated auditing, bias detection, and quality assurance tools to reduce noise and improve model performance.

Handshake appears to be positioning itself squarely within this higher-value segment.

The company was last valued at $3.3 billion in 2022 and reportedly expects to end 2025 with $300 million in annualized revenue run rate (ARR), with projections to reach the high hundreds of millions this year. It currently provides data services to eight leading AI labs, including OpenAI.

Industry observers see the Cleanlab acquisition as a logical next step for Handshake’s evolution.

Rather than competing solely on labor scale — a model often vulnerable to commoditization — Handshake is combining its long-standing talent supply network with research-driven data quality infrastructure. The strategy could give it a differentiated position relative to larger competitors such as Scale AI, which emphasizes full-stack data operations at enterprise scale, and quality-focused providers like Surge AI and expert marketplace platforms such as Mercor.

From Campus Recruiting to AI Infrastructure

Handshake’s transformation highlights a broader trend among HR and talent platforms seeking higher-growth adjacencies. What began as a university recruiting marketplace has gradually evolved into an AI data services provider, repurposing its core competency — matching specialized talent to complex tasks — for a new generation of customers.

With the addition of Cleanlab’s auditing technology, Handshake is no longer just supplying human labelers. It is building the systems to verify, measure, and continuously improve the quality of the data those humans produce.

In an AI economy where better data increasingly outweighs bigger datasets, that shift may prove decisive.