Deccan AI, a San Francisco-based startup with an army of talent based in Hyderabad, has raised $25 million. Its focus? Helping big tech keep their AI models sharp after the initial training phase.
The firm's services range from enhancing coding and agent capabilities to creating environments for reinforcement learning. With key customers like Google DeepMind and Snowflake on board, Deccan is navigating a rapidly growing market where quality data is king.
But the journey isn’t easy. Managing a network of 1 million contributors, many with advanced degrees, ensures accuracy but poses its own challenges. Reddy notes that while post-training work demands precision, it’s also time-sensitive, requiring quick turnaround times from massive datasets.
The sector is also facing scrutiny over working conditions and pay. Deccan offers a range of earnings, from $10 to $700 per hour, with top contributors earning up to $7,000 monthly. For now, India remains the key hub for talent, though the company is diversifying its sources.
As AI continues to shape our world, companies like Deccan are ensuring that these models behave as intended, making the tech more reliable and less prone to error.
