Sarvam AI's Young Team Achieves Breakthrough with 105B-Parameter LLM
In a remarkable feat of innovation, Sarvam, a Bengaluru-based generative AI company, has successfully built a foundational large language model (LLM) with 105 billion parameters. What makes this achievement extraordinary is that the core development was led by a compact team of just 15 researchers and engineers, many of whom are in their twenties and early thirties. This project highlights the burgeoning talent pool in India's AI sector and demonstrates how small, dedicated teams can tackle complex technological challenges.
Leadership and Key Contributors
The effort was spearheaded by Rahul Aralikatte, a postdoctoral researcher at the Mila - Quebec AI Institute, which is headed by Turing Award winner Yoshua Bengio. Aralikatte, who holds six US patents and has authored over 30 peer-reviewed papers with more than 1,000 citations, led the team's work across critical areas including data engineering, large-scale pre-training, evaluation frameworks, and safety guardrails for the model.
Among the key contributors were Sumanth Doddapaneni, a machine learning researcher at Sarvam and a fourth-year PhD student on leave from IIT Madras's computer science department. Doddapaneni managed the model's pre-training runs and contributed to data engineering. Other team members, such as Mohit Singla and Aashay Sachdeva, worked extensively on data engineering and post-training phases, while Anna Upreti focused on safety and alignment during the post-training stage.
Insights from Sarvam's Leadership
Vivek Raghavan, cofounder of Sarvam, expressed pride in the team's dedication. "The 105-billion-parameter model was trained completely from scratch by a very small team—about 15 people, most of whom are early in their careers. I sometimes jokingly call them 'kids', but they worked day and night to make it happen," he said. Raghavan emphasized the importance of identifying and nurturing high-potential talent, noting, "What matters is identifying the people who can become 10x or even 100x engineers."
Sarvam, which describes itself as a full-stack GenAI company, employs 40 technologists overall. The company's work spans the entire AI stack, from training core models from scratch and enhancing them through fine-tuning and reinforcement learning to developing harnesses and orchestration layers for practical applications.
Product Offerings and Talent Attraction
The company has launched several product families, including:
- Sarvam for Conversations: Enables AI-driven communication across India's diverse languages.
- Sarvam for Work: Focuses on enterprise workflows and internal process automation.
- Sarvam Studio: Designed to help create content in Indian languages and culturally relevant formats.
Additionally, Sarvam provides APIs for developers, with hundreds of companies and developers already leveraging them to build AI-powered products.
Raghavan highlighted two key factors that attract talent to Sarvam: "Many of them feel strongly about building technology for India's needs. And when you work at a company like Sarvam, you can contribute across a much larger scope of work. At a large multinational company, you might only work on a small piece of a much bigger system."
India's Rising AI Talent and Global Context
India's young professionals are rapidly emerging as AI natives, mirroring trends in the West where young founders and developers are leading cutting-edge AI startups. This shift is evident in companies like Mercor and Cursor, many of which are led by Indian-American entrepreneurs.
Commenting on India's AI talent, Manish Gupta, senior director at Google DeepMind, praised the innovation happening in the country. "We are inspired by how Indian innovators used these capabilities to build indigenous solutions. It is exciting to see local innovation thriving in India as we continue to advance the frontiers of general reasoning with Gemini and our other foundational models," he said. Gupta noted that teams in India are driving research to make AI models more efficient while better reflecting the country's linguistic and cultural diversity.
Full Team Roster
The dedicated team that worked across data engineering, pre-training, post-training, safety, and evaluation includes: Rahul Aralikatte, Sumanth Doddapaneni, Mohit Singla, Aashay Sachdeva, Anand Bollu, Harveen S, Arpit Dwivedi, Tanay Anand, Sushil Khyalia, Manav Singhal, Raghavan A K, Harshit Kedia, Ritvik Aryan Kalra, Anna Upreti, and Soham Petkar.
This achievement underscores the potential of India's tech ecosystem, where young, driven teams are pushing the boundaries of artificial intelligence and contributing significantly to the global AI landscape.
