Datacurve raises $15 million to fight Scale AI

As AI corporations have matured, the fight for high-quality data has turn into one of the best areas in the industry, giving rise to corporations reminiscent of Mercor, Surge and, most significantly, Alexander Wang’s Scale AI. But now that Wang has began leading artificial intelligence at Meta, many donors see it as an opportunity and are willing to fund corporations with compelling recent strategies for collecting training data.

Y Combinator graduate Data curve is one such company that focuses on high-quality data for software development. On Thursday, the company announced a $15 million Series A round led by Chemistry’s Mark Goldberg, with participation from employees of DeepMind, Vercel, Anthropic and OpenAI. The Series A follows a $2.7 million seed round in which former Coinbase CTO Balaji Srinivasan invested.

Datacurve uses a “bounty hunter” system to attract expert software engineers to complete the most difficult-to-obtain datasets. The company pays these contributions and has donated over $1 million in prizes to date.

But co-founder Serena Ge (pictured above with co-founder Charley Lee) says the biggest motivator is not financial. For high-value services reminiscent of software development, the pay for working with data will all the time be much lower than for conventional employment – so the most significant advantage for a company is a positive user experience.

“We treat this as a consumer product, not a data tagging operation,” Ge said. “We spend a lot of time thinking about: How can we optimize this so that the people we want are interested and come to our platform?”

This is especially essential as post-training data needs turn into increasingly complex. While earlier models were trained on easy datasets, today’s AI products rely on complex RL environments that should be constructed by collecting specific and strategic data. As environments turn into more sophisticated, data requirements turn into greater in each quantity and quality, which may provide an advantage for high-quality data collection corporations like Datacurve.

As an early-stage company, Datacurve focuses on software engineering, but Ge says the model might be just as easily applied to fields reminiscent of finance, marketing and even medicine.

“We are now building a post-training data collection infrastructure that attracts and retains highly competent people in their own fields,” Ge says.