Sarvam AI appoints Ex-Google Cloud India MD Sreedharan as president

Bengaluru: In its first high-profile C-suite hire, homegrown AI unicorn Sarvam AI has roped in former Google Cloud India MD Sashikumar Sreedharan as president and chief business officer, sources told TOI.Sreedharan recently stepped down from Google Cloud and is currently serving his garden leave period. Amit Kumar, managing director of Digital Natives Business at Google Cloud India, has been appointed interim MD, succeeding him.The appointment comes as Sarvam is building out its leadership bench and research capabilities even as it doubles down on its ambition to develop sovereign AI infrastructure and models in India. An email sent to Sarvam did not elicit a response till press time.The homegrown AI startup recently appointed AI researcher Devendra Chaplot, a founding member of Mistral AI and Thinking Machine Labs, as an adviser. California-based Chaplot is part of Sarvam’s efforts to tap global Indian AI talent.Sreedharan brings experience in building cloud businesses for enterprises and the public sector. At Google Cloud, he led the expansion of its sovereign cloud capabilities in India, catering to organisations with stringent requirements around data security, control and isolation. He was previously managing director of Microsoft India and CEO and MD of Happiest Minds.Sarvam’s leadership expansion follows a $234 million funding round led by HCLTech and Bessemer Venture Partners that valued the company at $1.5 billion, making it one of India’s newest AI unicorns. The Bengaluru-based startup’s revenue was Rs 45 crore in FY26, according to unaudited financials.The company is now planning to build a trillion-parameter AI model and establish a US office to attract Indian-origin AI researchers. At the company’s recent flagship event, its founder Pratyush Kumar told TOI the model will target capabilities in coding, cybersecurity, scientific research and simulation, bringing Sarvam closer to the frontier of global AI models.The company’s current 100-billion-parameter model is already being used for voice AI, conversational systems, English speech recognition and dictation, besides more complex tasks, Kumar said in a recent interaction with TOI. Sarvam has so far processed 325 million minutes of voice calls, he told TOI.“India should be producing the AI tokens it consumes,” Kumar told TOI, arguing that continued dependence on overseas models could leave India paying not only for access to AI but also through the data generated by its users.Raghavan said Sarvam wants to build a “real token factory in India”. Serving AI tokens from within the country, he said, would be “the first step to becoming sovereign”.The company also believes sovereignty cannot come at the cost of performance. “Sovereignty should not be a tax,” Kumar said, arguing that Indian AI products must be benchmarked against the best global models rather than getting a pass simply because they are homegrown.Sarvam is already building its model portfolio. Its latest speech recognition model, Saaras V4, combines an audio encoder with a 3-billion-parameter hybrid state-space model decoder, trained entirely in-house from scratch.Sarvam said Saaras V4 delivers state-of-the-art performance across all 22 Indian languages, including low-resource languages. It also records the lowest word error rate across seven global English datasets, covering six non-native English dialects and Indian English.You use AI every day. Now get your AI Quotient. Take the AIQ test.