Deccan AI Secures $25 Million in Series A Funding
Deccan AI, a promising startup that specializes in post-training data and evaluation work for Artificial Intelligence (AI) models, has successfully raised $25 million in its first major funding round. This achievement underscores the increasing demand for training and fine-tuning AI models. Deccan AI has entrusted a significant portion of this crucial work to an expert workforce based in India.
The Series A all-stock funding round was spearheaded by A91 Partners and saw participation from Susquehanna International Group and Prosus Ventures.
Addressing the Demand for Post-Training AI Work
While organizations like OpenAI and Anthropic construct basic AI models in-house, a significant part of the post-training work, including data generation, evaluation, and reinforcement learning, is typically outsourced. This practice is becoming more prevalent as companies strive to ensure the reliability of their systems in real-world applications. Deccan AI is a forefront entity among new startups that are responding to this demand.
Established in October 2024, Deccan offers services that range from enhancing coding and agent capabilities of models to training systems to interact with external tools like application programming interfaces (APIs). APIs are crucial in connecting AI models to software systems.
Notable Contributions and Clients
Deccan AI collaborates with pioneering labs on tasks such as generating expert feedback, conducting assessments, and creating reinforcement learning environments. The startup also serves businesses through its products such as the assessment suite Helix and an operations automation platform. As AI models continue to evolve beyond text-based models to “world models” that can better comprehend physical environments, including robotics and vision systems, the work of Deccan AI also adapts accordingly.
Deccan AI has an impressive client roster, including tech giants like Google DeepMind and Snowflake. According to Rukesh Reddy, the founder of Deccan AI, the startup currently manages about 20 active projects and has onboarded approximately 10 clients.
Deccan AI’s Workforce
With its headquarters in the San Francisco Bay Area and a large operations team situated in Hyderabad, India, Deccan AI employs around 125 individuals. The startup also leverages a network of over a million contributors, including students, domain experts, and PhDs. Reddy revealed that about 5,000 to 10,000 contributors are active in a typical month.
Reddy further divulged that approximately 10% of Deccan’s contributors possess higher degrees such as a master’s or doctorate. However, this percentage could be higher among active contributors, depending on the project requirements.
Competitive Landscape and Challenges
The market for AI training services has seen rapid growth, especially with the rise of large language models. Companies like Meta-owned Scale AI and rival Surge AI, as well as startups Turing and Mercor, are competing to provide data labeling, assessment, and reinforcement learning services.
Reddy acknowledged that maintaining quality remains a challenge in this sector. Post-training work requires very precise, domain-specific data that is difficult to scale due to its complexity. Errors after training can significantly impact model performance in production, necessitating near-zero error tolerance.
The sector also faces criticism over working conditions and wages. However, Reddy assured that earnings on the Deccan platform range between $10 and $700 per hour, with top contributors earning up to $7,000 per month.
India as a Hub for AI Training Talent
Most of Deccan AI’s contributors are based in India, despite the majority of their clients being US-based AI labs. This is a strategic decision by Deccan AI to concentrate most of its workforce in India in order to better manage quality. Reddy stated, “If you have operations in a single country, it becomes much easier to maintain quality.”
This strategy highlights India’s current position in the global AI value chain, primarily as a provider of talent and training data. However, Reddy mentioned that Deccan has begun scouting for talent in other markets, including the US, to acquire niche expertise in geospatial data and semiconductor design.
Deccan AI, designed as a “born GenAI” company, focused on more skilled work from the start. The startup has grown 10-fold in the past year and is currently at a double-digit revenue run rate of $1 million. About 80% of its revenue comes from its top five customers, reflecting the concentrated nature of the frontier AI market.
For more details, click here.
