Deccan AI, a fast-rising startup specialising in post-training data and evaluation for artificial intelligence models, has secured $25 million in its first major funding round, underscoring growing investor confidence in a critical but often overlooked segment of the AI value chain.
The all-equity Series A round was led by A91 Partners, with backing from Susquehanna International Group and Prosus Ventures.
The investment comes at a time when demand for refining and stress-testing AI systems is accelerating, as companies race to deploy models that can operate reliably in real-world environments.
According to founder and chief executive Rukesh Reddy, the company currently serves around 10 clients and manages several dozen active projects at any given time.
“Many of our competitors go to 100-plus countries to find the experts,” he said. “If you have operations in just one country, it becomes far easier to maintain quality,” said Rukesh Reddy.
What you need to know
While leading AI developers such as OpenAI and Anthropic continue to build foundational models in-house, a significant portion of the post-training process is increasingly being outsourced.
This stage, which includes data generation, evaluation, and reinforcement learning is essential to improving model accuracy, safety, and usability.
As expectations for AI performance rise, so too does the need for highly specialised, human-driven input.
Deccan AI is part of a new wave of companies positioning themselves at the centre of this shift, providing the expertise and infrastructure required to fine-tune advanced systems.
Building capabilities beyond text-based AI
Founded in October 2024, Deccan AI offers a range of services designed to enhance model performance across multiple domains.
These include improving coding and agent-based capabilities, as well as enabling AI systems to interact with external tools such as application programming interfaces (APIs), which allow models to connect with broader software ecosystems.
The company also works with frontier AI labs on complex tasks such as generating expert feedback, conducting evaluations, and developing reinforcement learning environments.
Its product suite includes Helix, an evaluation platform, alongside an operations automation tool aimed at enterprise clients.
Notably, the company is already adapting to the next frontier of AI development, so-called “world models” that extend beyond text into physical environments, including robotics and computer vision.
Blue-chip clients and rapid growth
Deccan AI counts major technology players such as Google DeepMind and Snowflake among its customers.
Headquartered in the San Francisco Bay Area, with a significant operational base in Hyderabad, the company employs approximately 125 staff and leverages a vast global contributor network exceeding one million individuals.
Between 5,000 and 10,000 contributors are active each month, spanning students, domain experts, and PhD holders. Around 10% of the broader contributor base holds advanced degrees, with higher concentrations on more technically demanding projects.
This hybrid model, combining a lean core team with a large, flexible workforce has enabled Deccan AI to scale rapidly. The company reports 10-fold growth over the past year and has reached a double-digit million-dollar revenue run rate, although exact figures were not disclosed.
A competitive and fast-evolving market
The market for AI training and evaluation services has expanded in tandem with the rise of large language models, attracting a growing number of competitors.
Established players such as Scale AI and Surge AI, alongside newer entrants like Turing and Mercor, are all vying for a share of the rapidly growing sector.
However, Deccan AI is attempting to differentiate itself by focusing on higher-skill, post-training tasks rather than traditional data labelling.
Reddy said that the work is highly time-sensitive, with AI labs often requiring large volumes of accurate, domain-specific data within tight deadlines, sometimes in a matter of days.
What this means
As with much of the AI training industry, the sector has faced scrutiny over working conditions and compensation, particularly given its reliance on large pools of contract workers.
Reddy said contributors on Deccan’s platform earn between $10 and $700 per hour, with top performers making up to $7,000 per month, depending on expertise and project complexity.
Unlike traditional data labelling firms that originated in computer vision tasks, Deccan AI was built from the ground up as a generative AI-focused company.
This “born GenAI” approach has allowed it to concentrate on higher-value, complex work from the outset, a positioning that could prove advantageous as AI systems become more sophisticated.
As the AI arms race intensifies, Deccan AI is betting that its focus on quality, speed, and specialised talent will secure its place as a key enabler of the next generation of intelligent systems.
Talking Points
It is striking that Deccan AI is building its business around one of the most critical yet underreported layers of artificial intelligence, post-training, where models are refined, tested, and made reliable for real-world use.
This positioning places the company at the heart of a fast-growing demand cycle, as AI labs increasingly outsource complex tasks such as evaluation, reinforcement learning, and expert data generation to improve model performance.
At Techparley, we see this as a clear signal that the AI value chain is expanding beyond model creation, with significant opportunities emerging in the infrastructure and services that make these systems usable at scale.
Deccan’s focus on higher-skill work, rather than traditional data labelling, suggests a deliberate move up the value chain, one that could offer stronger margins and long-term defensibility in an increasingly crowded market.
With fresh capital and rapid growth, Deccan AI is well placed to become a key enabler in the global AI ecosystem but its long-term success will depend on how effectively it balances quality, scale, and evolving industry expectations.
——————-
Bookmark Techparley.com for the most insightful technology news from the African continent.
Follow us on Twitter @Techparleynews, on Facebook at Techparley Africa, on LinkedIn at Techparley Africa, or on Instagram at Techparleynews.
