Handshake acqui-hires Cleanlab to raise data-label quality for AI labs

Training data quality is becoming a competitive moat for model builders, and Handshake is making a talent-heavy bet to improve it. The hiring platform turned data-labeling provider has acquired Cleanlab to bring automated label-auditing research in-house for AI data production.

Acqui-hire adds MIT PhDs to Handshake’s data-labeling research

Handshake says the transaction is primarily an acqui-hire: nine key Cleanlab employees are joining its research organization, including co-founders Curtis Northcutt, Jonas Mueller, and Anish Athalye, all MIT computer science PhDs. Terms were not disclosed.

Cleanlab, founded in 2021, built software designed to catch labeling errors without needing a second human reviewer. In practice, these systems score or flag “likely wrong” examples so teams can fix them faster—useful when projects involve domain experts and expensive annotation cycles.

For B2B marketers and e-commerce founders, the direct takeaway is downstream: better training data tends to reduce hallucinations and edge-case failures in customer-facing automations (support agents, product classification, compliance checks), which can translate into fewer escalations and more stable workflows.

Why label-auditing matters as data vendors converge

Handshake started in 2013 as a platform for hiring college graduates and launched a human data-labeling business roughly a year ago, serving foundational model companies. It has provided data to eight top AI labs, including OpenAI.

Cleanlab raised 30 million dollars from investors including Menlo Ventures, TQ Ventures, Bain Capital Ventures, and Databricks Ventures, and previously scaled to more than 30 employees.

Northcutt said Cleanlab saw interest from other labeling firms, but opted for Handshake because competitors often rely on Handshake’s marketplace to source specialized human experts (doctors, lawyers, scientists) for annotation work.

Handshake was last valued at 3.3 billion dollars in 2022 and was forecast to end 2025 at 300 million dollars in annualized revenue run rate. It is reportedly tracking toward “high hundreds of millions” in ARR this year, according to reporting by Upstarts Media.

The strategic signal: as labeling becomes more commoditized, auditing and quality control are moving to the center—especially for regulated or high-stakes datasets where one bad label can poison model behavior.