(left) Sravanth Aluru, CEO, and Gaurav Baid, CPO of Avataar AI
For Aluru, who has an engineering background from IIT Bombay and experience at Microsoft before studying at Wharton, this was not merely a passing trend. He perceived it as the start of a technological revolution akin to the internet. “Intelligence has never been challenged before,” he remarked. “Suddenly, we were discussing a future where substantial parts of human intelligence in the workplace could be enhanced by an assistant.”
This vision influenced him when he co-founded Avataar AI in 2015. Unlike companies like OpenAI focusing on artificial general intelligence, Aluru aimed to create AI solutions that are practical, cost-effective, and applicable to real-world problems. Presently, Avataar AI is one of eight startups selected under the Government of India’s IndiaAI Mission to develop large language models (LLMs).
Avataar AI distinguishes itself by concentrating on task-specific models that are smaller and more focused on particular applications such as agriculture, finance, or small-to-medium enterprise (SME) workflows. Aluru compares massive trillion-parameter models to asking an Einstein to solve an algebra problem, emphasizing the importance of efficiency through smaller, specialized models. The company employs distillation and reinforcement learning to enhance accuracy while reducing operational costs, leading to models that are economically viable and approach near-perfect accuracy in their designated tasks.
India AI Mission
Utilizing trillion-parameter models for specific tasks can be financially impractical. Avataar AI uses expansive general-purpose models as “Teacher Models” to extract domain knowledge into more efficient “Student Models.” These specialized models undergo supplementary refinement through reinforcement learning. This approach mirrors the educational process where a professor conveys complex ideas once, distilling them into tailored lessons for each student. Consequently, the smaller models gain essential knowledge specific to their domain, thereby enhancing accuracy while significantly lowering costs.
Through the IndiaAI Mission, the startup has access to 768 NVIDIA GPUs in its initial phase, alongside government-enhanced infrastructure and financial assistance. Over the following six months, Avataar AI intends to showcase its inaugural workflows on AI Cache, which will include farmer advisory systems integrating plant disease detection with satellite weather data, and enterprise solutions such as automated invoicing, video generation for product listings, and contract management.
Funds raised
To date, Avataar AI has secured $45 million in funding from Sequoia Capital and Tiger Global, establishing a client base in retail and healthcare across the U.S. With the support of the IndiaAI Mission, the company is restructuring its Bengaluru headquarters to serve as a central hub for developing scalable, efficient AI solutions. Specifically for the IndiaAI Mission, Avataar is hosting its Agentic platform and select specialized workflows on AIKosh, aiming to democratize AI access for citizens, SMEs, enterprises, and government bodies. These services will be available to all, supported by government-subsidized computing, with a pay-per-use model at significantly reduced costs compared to current cloud service providers.
For Aluru, the IndiaAI Mission signifies more than a business opportunity. “But here, there is a greater purpose: making AI accessible to citizens, SMEs, and the government. If we can contribute to that, I think we’ll complete this journey with the biggest smile.”
Published on October 13, 2025
