As the demand for training and improving AI models increases, IA Deccan – a startup providing post-training data and assessment work – has raised $25 million in its first major funding round, with much of this work carried out by an expert workforce based in India.
The all-stock Series A round was led by A91 Partners, with participation from Susquehanna International Group and Prosus Ventures.
While cutting-edge AI labs including OpenAI and Anthropic build basic models in-house, much of the post-training work – from data generation to evaluation and reinforcement learning – is increasingly outsourced as companies work to make systems reliable in real-world use. Deccan is part of a new set of startups responding to this demand.
Founded in October 2024, Deccan provides services ranging from helping models to improve coding and agent capabilities to training systems to interact with external tools such as application programming interfaces (APIs), which connect AI models to software systems.
The startup works with pioneering labs on tasks such as generating expert feedback, conducting assessments, and creating reinforcement learning environments, while also serving businesses through products such as its assessment suite, Helix, and an operations automation platform. The work also evolves as models move beyond text to become “world models” that better understand physical environments, including robotics and vision systems.
Deccan’s clients include Google DeepMind and Snowflake, according to the company. It has onboarded about 10 clients and runs about 20 active projects at any given time, founder Rukesh Reddy (pictured above) said in an interview.
The startup, headquartered in the San Francisco Bay Area and with a large operations team in Hyderabad, employs around 125 people and leverages a network of over a million contributors, including students, domain experts and PhDs. About 5,000 to 10,000 contributors are active in a typical month, Reddy told TechCrunch.
Techcrunch event
San Francisco, California | October 13-15, 2026
About 10% of Deccan’s contributors have higher degrees such as master’s and doctorate, although the share is higher among active contributors depending on the project requirements, Reddy said.
The AI training services market has developed rapidly alongside the rise of major language models, with companies such as Scale AI belonging to the meta and his rival AI emergesas well as startups Turing And Mercor competing to provide data labeling, rating, and reinforcement learning services.
“Quality remains an unsolved issue,” Reddy said, adding that error tolerance after training is “close to zero” because errors can directly affect model performance in production. This makes post-training more complex than previous steps, requiring very precise, domain-specific data that is more difficult to scale.
The work is also very time-sensitive, he said, with AI labs sometimes requiring large volumes of high-quality data in a matter of days, making it difficult to balance speed and accuracy.
The sector has faces criticism over working conditions and salarieswith large pools of on-demand workers often used to generate training data. Reddy said earnings on the Deccan platform range between $10 and $700 per hour, with top contributors earning up to $7,000 per month.
India emerging as a hub for AI training talent
Although its clients are largely US-based AI labs, most of Deccan’s contributors are based in India. Competitors such as Turing and Mercor also find contractors of the country, but operate through a broader set of emerging markets.
Deccan chose to concentrate much of its workforce in India to better manage quality, Reddy said. “Many of our competitors go to over 100 countries to find experts,” he said. “If you have operations in a single country, it becomes much easier to maintain quality. »
This approach highlights India’s current position in the global AI value chain – as a provider of talent and training data rather than a developer of frontier models, which remain concentrated among a handful of US companies and a few players in China.
However, Reddy said Deccan has started looking for talent in a few other markets, including the US, to gain niche expertise in geospatial data and semiconductor design.
Reddy said Deccan was designed as a “born GenAI” company, unlike traditional data labeling companies that started with computer vision tasks. This means that he focused on more skilled work from the start.
Deccan has grown 10-fold in the past year and is now at a double-digit revenue run rate of $1 million, Reddy said, declining to share details. About 80% of its revenue comes from its top five customers, reflecting the concentrated nature of the frontier AI market, it added.