Tech Mahindra launches AI hub to propagate Hindi-language AI model

Located within Tech Mahindra’s Makers Lab, the site was set up with the intention of bringing GenAI, LLMs, and digital twins into mainstream use in the region and is powered by Nvidia’s AI Enterprise software platform.

The main vector of this plan appears to be Project Indus 2.0 – an LLM run on Nvidia NeMo which is based on Hindi and dozens of its dialects, such as Bhojpuri, Dogri, and Maithili, and is aimed at sectors such as retail, banking, healthcare, and citizen services in India.

The CoE has also tapped up the Nvidia Omniverse platform for the development of industrial AI digital twins and ‘physical AI applications’ aimed at various sectors, including manufacturing, automotive, telecommunications, healthcare, banking, financial services and insurance. 

“At Tech Mahindra, we are redefining the boundaries of AI innovation,” said Atul Soneja, Chief Operating Officer, Tech Mahindra. “Collaborating with NVIDIA, we are setting a new benchmark for enterprise-grade AI development by seamlessly integrating GenAI, industrial AI and sovereign large language models into the heart of global enterprises and industries.”

John Fanelli, Vice President, Enterprise Software at Nvidia added: “Built with NVIDIA technology, Tech Mahindra’s Center of Excellence will accelerate the development and adoption of sovereign AI LLMs and applications tailored for India’s diverse industries and linguistic landscape. Our collaboration with Tech Mahindra is helping create a foundation for AI-driven innovation in the region and around the world.”

Going forward the intention for Indus 2.0 is to include agentic workflows and support multiple dialects to provide a ‘more nuanced and effective AI solution tailored to India's diverse linguistic and cultural landscape.’ Agentic AI ‘significantly improves’ productivity by enabling AI applications to learn, reason, and take action, we’re told.