As demand for training and refining AI models grows, Deccan AI — a startup that provides post-training data and testing work — has raised $25 million in its first round of funding, with much of that work being done by a team of professionals based in India.
The all-equity Series A round was led by A91 Partners, with participation from Susquehanna International Group and Prosus Ventures.
While frontier AI labs including OpenAI and Anthropic build valuable models in-house, much of the post-training work — from data generation to testing and reinforcement learning — is increasingly outsourced as companies push to make systems more reliable for real-world use. Deccan is emerging as one of a new set of startups that are capitalizing on that need.
Founded in October 2024, Deccan provides services from helping models develop coding skills with agents to training programs to interact with external tools such as application programming interfaces (APIs), which connect AI models and software systems.
The startup works with frontier labs on tasks such as generating expert feedback, conducting testing and building dynamic learning environments, while also helping businesses with products including its testing environment, Helix, and automated work platform. The work is also evolving as models move beyond text to so-called “world models” that better understand physical environments, including robotics and vision systems.
Deccan’s customers include Google DeepMind and Snowflake, according to the company. It has logged about 10 clients and runs several active projects at any given time, founder Rukesh Reddy (pictured above) said in an interview.
The startup, headquartered in the San Francisco Bay Area with a large operations team in Hyderabad, employs about 125 people and relies on a network of more than 1 million contributors, including students, domain experts, and PhDs. About 5,000 to 10,000 participants are active in a typical month, Reddy told TechCrunch.
Techcrunch event
San Francisco, CA
|
October 13-15, 2026
About 10% of the Deccan donor base has advanced degrees such as master’s and PhDs, although the share is larger among active donors depending on project needs, Reddy said.
The market for AI training services has grown rapidly in line with the rise of large-scale linguistic models, with companies such as Meta-owned Scale AI and its competitor Surge AI, as well as startups Turing and Mercor competing to provide data labeling, evaluation, and reinforcement learning services.
“Quality remains an unsolved problem,” said Reddy, adding that error tolerance after training is “close to zero” since errors can directly affect the model’s performance in production. That makes post-training more difficult than earlier stages, requiring more precise, domain-specific data that is difficult to measure.
The work is also very time-sensitive, he said, with AI labs sometimes needing large volumes of high-quality data within days, making it difficult to accurately measure speed.
The sector has faced criticism over working conditions and pay, with large pools of gig workers often used to generate training data. Reddy said earnings in the Deccan range from $10 to $700 an hour, with top contributors earning up to $7,000 a month.
India is emerging as a talent hub for AI training
Even though its customers are mainly US-based AI labs, most of Deccan’s donors are based in India. Competitors such as Turing and Mercor also source contractors from the country, but they operate in all emerging markets.
Deccan chose to focus more on its employees in India to better manage quality, Reddy said. “Many of our competitors go to more than 100 countries to find experts,” he said. “When you work in just one country, it’s much easier to maintain the standard.”
That approach highlights India’s current position in the global AI value chain — as a supplier of talent and training data rather than a developer of frontier models, which remain focused on a few US companies and a few players in China.
However, Reddy said Deccan has started sourcing talent from several other markets, including the US, for niche expertise in geospatial data and semiconductor design.
Reddy said Deccan was built as a “GenAI-born” company, unlike traditional data labeling companies that started with computer vision jobs. This means it focuses on high-skilled work from the start.
Deccan has grown 10x in the past year and is now at double-digit dollar spending levels, Reddy said, declining to share directly. About 80% of its revenue comes from its top five customers, reflecting the concentrated nature of the frontier AI market, he added.
