DataCurve's 18-year-old founder raises $15M Series A to supply high-skill coding data to frontier AI labs

Summary

Serena, founder of DataCurve, closed a $15 million Series A that brings total funding to $17.7 million. She launched the company at 18 after pivoting out of Y Combinator, where she entered with an unrelated web-agent concept. DataCurve has been operating for roughly a year.

The business supplies high-skill coding data to frontier AI labs for post-training, covering both supervised fine-tuning and reinforcement learning workflows. The core thesis is that current coding agents can handle roughly two-hour tasks, and labs need purpose-built training data to push that ceiling toward seven-hour or longer horizon tasks, including intermediate verifiers and multimodal data science pipelines.

DataCurve positions itself explicitly against the commodity end of the data labeling market. Rather than recruiting niche domain experts or running fully automated RL pipelines, it targets elite software engineers through a bounty-based system, arguing that long-term contracts produce low-effort work from coasters billing $200 an hour. The platform has paid out $1 million in bounties to date.

The product strategy centers on developer experience and pipeline tooling, including gamification, to retain high-skill contributors through complex, multi-hour tasks. Serena draws a clear distinction from broker-model competitors: DataCurve owns the full pipeline internally and sells finished data to labs, not raw labor or introductions.

First commercial traction came via inbound interest from a FAANG-tier customer before Serena had ever run a sales call, a signal she cites as validation of genuine demand rather than outbound hustle. The raise and platform milestone suggest the labs-as-customer model is scaling, though DataCurve remains a young, single-vertical business in a market where Scale AI and automated RL pipelines are direct competitive pressures.