Prof.ai to founder.ai

When the Covid-19 pandemic struck in 2020, Sashikumaar Ganeshan and his colleagues at Indian Institute of Science, Bangalore were tasked with a project; to build a model to predict the spread of the contagion. For Ganeshan, a professor who spent two decades in computational mathematics in premier academic institutions, that was an epiphany.

The work on Covid prediction model was his deep immersion into AI technologies. “I have gone into Python, and machine learning neural networks. I want to bring this technology into science and engineering and that is how the research started,” Ganesan said.

Ganesan, who currently heads the AI for Research and Engineering eXcellence (AiREX) – at IISc, is focused on using AI to tackle scientific and engineering issues across sectors such as manufacturing through his startup, ZenteiQ.ai. With the startup, Ganesan and his team are building scientific foundational model Brahm AI and is backed by the IndiaAI mission.


The pattern

Ganeshan’s path from computational research to entrepreneurship is not unusual but a familiar pattern in AI startups. Creation and commercialisation of intellectual property have made rigorous technology research central to building AI companies. Just as Ganeshan built Zenteiq, Arjun Jain, former researcher at Apple and adjunct faculty at the Indian Institute of Sciences, founded Fastcode.ai. Prathosh AP, co-founder, Latentforce.ai, is a faculty at IISc in machine and deep learning. Amritendu Mukherjee, researcher and former faculty at IISc cofounded NeuroPixel.ai which he eventually exited.

Before starting Sarvam, Pratyush Kumar was a researcher at Microsoft, and faculty at the Indian Institute of Technology, Madras. Vyas Sekar, professor at Carnegie Mellon University cofounded Rockfish.ai.

Researchers starting up is an established global trend in AI company building. Anthropic was founded by former OpenAI researchers and Cohere was founded by NLP researchers and former Google researchers.

Ashish Vaswani and Niki Parmer, former Google researchers, started Essential AI; Arthur Mensch, founder of Mistral AI is a former DeepMind scientist. Fei-Fei Li, computer science professor at Stanford founded World Labs, Yann LeCun, former researcher at Meta founded Advanced Machine Intelligence, and Andrew Ng, adjunct professor at Stanford, founded Coursera and DeepLearning.

Researcher’s moment

This trend is gathering pace and is increasingly visible in India’s fledgling AI startup landscape. There is a reason. Sanjeev Kumar, advisor - AI and data infrastructure at Avataar Venture Partners, said a large percentage of technologists who have been passionately pursuing AI research, either within the tech industry or in academia, are now seeing commercialisation opportunities. “This phenomenon has been accelerating in the last couple of years because there is potential for AI research to be commercialized,” he said.

Another aspect is the inclination of investors to fund such companies because their IP is unique and defensible, he added. In addition, OpenAI, Anthropic and Google DeepMind are sources of research talent for many of these AI startups, who are now racing to get ahead in the AI arms race.

World Labs, led by Fei-Fei Li raised $230 million at a $1 billion valuation in 2025. Sarvam raised $50 million cumulatively in 2025, one of the largest early funding rounds the segment has seen in India.

A US-based AI researcher, who is starting up, said, "If you are a researcher, instead of working on a large firm, starting up gives you the best opportunity to create intellectual property and also to create groundbreaking research that is now possible,” he added.

Apoorva Pandhi, Managing Partner at Zetta Venture Partners, noted that AI researchers add the most leverage at the infrastructure and data layers, where meaningful technical breakthroughs are still required.

Foundational knowledge

Aravind Jeyendran, founder & CEO, Latentforce.ai, pointed out that unlike software development, to understand machine learning one needs to understand how LLMs work.

Say for example, I'm using AI for code generation, and I know that there are maybe ‘n’ number of problems that I have to solve within code generation. There should be somebody who understands these models, and understand that for problem A, you can use LLMs, problem B, you cannot use LLMs and for Problem C, LLMs in the current form can’t be used and you need to use augmentation on it.

“Only a researcher who has worked a lot with these LLMs knows why. Because, figuring out what behaviour an LLM supports, or what you can do comes only if you have worked a lot with an LLM, and those are typically researchers,” he said.

In addition, founders also pointed out that only AI researchers, who have been working with LLMs and deep learning models, understand the trajectory, which is necessary to stay two steps ahead for product development and get a right to win. “If you set a product based on the current state of models and say that I will build this right now, you will fail,” Jeyendran said.

Challenges

One of the challenges is the concerns around how capable researchers are when it comes to commercialisation. Researchers are obsessed with the technical superiority of their product but they aren’t necessarily equipped with skills for distribution, partnerships, pricing and positioning.

“Research is long term. But if you are starting the company you have to make revenue and not just spend time in research for 10 years. Most researchers might not think like that,” one of the founders said, seeking anonymity.

A US-based investor, on the condition of anonymity said, one of the red flags is that many researchers continue to be professors, which might not bode well in the long term. “If you're a full-time faculty and you're a founder of a company, investors won't touch you because there's a significant equity that's locked up by a guy who's not invested full time,” he pointed out.

Gokul NA, co-founder, CynLr Robotics, highlighted that one of the challenges is also lack of AI research ecosystem that can produce quality researchers like that of the US.

But things are changing. “There are certain venture capitalists and organizations that support researchers who build companies for the sake of research. Even OpenAI was built like that,” said the founder, quoted earlier.

More researchers are now taking up entrepreneurship such as Ganesh Ramakrishnan, heading Government-backed BharatGen and Cyran AI solutions, led by Manan Suri, professor from the IIT Delhi.