As artificial intelligence companies evolve, the competition for premium data has intensified, making it one of the industry’s fiercest battlegrounds. This has propelled companies such as Mercor, Surge, and especially Alexandr Wang’s Scale AI into the spotlight. With Wang now leading AI efforts at Meta, many investors believe there is an opportunity to back startups with innovative methods for gathering training data.
Datacurve, a Y Combinator alum, is one such startup, concentrating on delivering top-tier data for software engineering. On Thursday, the company revealed it has secured $15 million in Series A funding, with Mark Goldberg from Chemistry heading the round and support from employees at DeepMind, Vercel, Anthropic, and OpenAI. This follows a $2.7 million seed investment, which included backing from former Coinbase CTO Balaji Srinivasan.
Datacurve has adopted a “bounty hunter” approach, inviting talented software developers to tackle the most challenging datasets. Contributors are compensated for their efforts, and the company has already distributed more than $1 million in bounties.
However, co-founder Serena Ge (pictured above with co-founder Charley Lee) emphasizes that money isn’t the main incentive. For specialized work like software engineering, the financial rewards for data tasks are always much less than traditional jobs—so the company’s key differentiator is creating a rewarding user experience.
“We see this as a consumer-facing product, not just a data labeling service,” Ge explained. “We invest significant effort into figuring out how to make our platform appealing and engaging for the people we want to attract.”
This focus is increasingly crucial as the complexity of post-training data requirements grows. While earlier AI models relied on straightforward datasets, modern AI systems depend on intricate RL environments that must be carefully built through targeted data collection. As these environments become more advanced, the demand for both higher quality and greater quantities of data increases—a trend that could benefit specialized data providers like Datacurve.
Currently, Datacurve is concentrating on software engineering, but Ge believes their approach could be extended to other sectors, including finance, marketing, and healthcare.
“Our current mission is to build an infrastructure for post-training data collection that draws in and keeps highly skilled professionals within their respective fields,” Ge says.
