As demand grows for training and refining AI models, Deccan AI — a startup supplying post-training data and evaluation work — has raised $25 million in its first major funding round. Much of the company's work is carried out by a large workforce of experts based in India.
The all-equity Series A round was led by A91 Partners, with participation from Susquehanna International Group and Prosus Ventures.
Founded in October 2024, Deccan provides services ranging from helping models improve coding and agent capabilities to training systems to interact with external tools such as APIs. The startup works with frontier AI labs on tasks like generating expert feedback, running evaluations, and building reinforcement learning environments.
Big Names on the Client List
Despite being less than two years old, Deccan has already secured some impressive clients. The company's customers include Google DeepMind and Snowflake. It has onboarded about 10 customers and runs a couple of dozen active projects at any given time, according to founder Rukesh Reddy.
Deccan grew 10x over the past year and is now at a double-digit million-dollar revenue run rate. However, Reddy declined to share exact figures. About 80% of the company's revenue comes from its top five customers, reflecting the concentrated nature of the frontier AI market.
The startup also serves enterprises through products including its evaluation suite called Helix and an operations automation platform. The work is also evolving as models move beyond text into so-called "world models" that better understand physical environments, including robotics and vision systems.
A Million Contributors, Mostly From India
Deccan is headquartered in the San Francisco Bay Area with a large operations team in Hyderabad. The company employs about 125 people and relies on a network of more than one million contributors, including students, domain experts, and PhDs. Around 5,000 to 10,000 contributors are active in a typical month.
About 10% of the contributor base holds advanced degrees such as master's and PhDs, though the share is higher among active contributors depending on project requirements.
Deccan chose to concentrate much of its workforce in India to better manage quality. Reddy explained that many competitors go to over 100 countries to find experts, but having operations in just one country makes it far easier to maintain quality.
This approach highlights India's current role in the global AI value chain — as a major supplier 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, Deccan has begun sourcing talent from a few other markets, including the US, for niche expertise in geospatial data and semiconductor design.
The Quality Challenge in Post-Training
The AI training services market has expanded rapidly alongside the rise of large language models, with companies like Scale AI, Surge AI, Turing, and Mercor all competing in the space. But Deccan believes quality is what sets it apart.
Reddy said that quality remains an unsolved problem in post-training, adding that tolerance for errors is close to zero as mistakes can directly affect model performance in production. This makes the work more complex than earlier stages of AI development, requiring highly accurate, domain-specific data that is difficult to scale.
The work is also highly time-sensitive, with AI labs sometimes requiring large volumes of high-quality data within days, making it challenging to balance speed with accuracy.
Fair Pay and Worker Conditions
The AI training industry has faced criticism over working conditions and low pay for gig workers who generate training data. Reddy said earnings on Deccan's platform range from about $10 to $700 per hour, with top contributors earning up to $7,000 a month.
These figures suggest Deccan is positioning itself at the higher end of the pay scale compared to traditional data labeling operations, which aligns with its focus on expert-level, high-accuracy work rather than bulk data annotation.
Built for the GenAI Era
Reddy said Deccan was built as a "born GenAI" company, in contrast to traditional data labeling firms that began with computer vision tasks. This means the startup has focused on higher-skill work from the very beginning — evaluating language model outputs, building reinforcement learning pipelines, and providing the kind of nuanced expert feedback that frontier AI labs need to make their models reliable in the real world.
With $25 million in fresh funding and a rapidly growing revenue base, Deccan AI is well-positioned to become a critical link in the AI supply chain — connecting India's deep pool of technical talent with the world's most advanced AI laboratories.
