Kedar Kulkarni wants his company to outlive him

A startup pantry is an inconvenient place to ask what one ought to do with one’s life. There are usually more immediate questions involving lunch, deadlines and the bank balance. Kedar Kulkarni and his colleagues asked anyway. Eleven people, a company with roughly six months of runway, and a question large enough to make the room feel smaller: what work would they choose if money and society’s expectations ceased to interfere?

Education and rural employment were among the answers. HyperVerge was then trying to find its footing with consumer photo applications. The distance between those ambitions and the product in their hands was becoming difficult to ignore. A company can possess clever technology and still struggle to explain why it deserves another year.

That conversation gives Kulkarni’s career its particular shape. Today, the Bengaluru-based co-founder and CEO runs a business selling identity verification, customer onboarding and fraud prevention. He also wants to help build at least 100 businesses that combine commercial endurance with community contribution. The ambition has a practical condition attached: someone must keep paying the bills.

A metallurgy student teaches machines to see

Kulkarni studied at IIT Madras from 2008 to 2012, graduating in metallurgy and materials engineering. His extracurricular interests proved unusually industrious. Alongside his degree, he helped establish the Computer Vision Group and co-founded Lema Labs, which taught students emerging technologies. Education and computer vision were already sharing his attention before either became a line in a company biography.

The robotics team at the institute’s Centre for Innovation offered an opening. As a first-year student, he was less experienced than his teammates. But their autonomous robot needed to interpret its surroundings, and computer vision was territory he could explore. A machine that cannot distinguish one room from another has limited prospects as an independent traveller.

He liked the immediacy: work through a problem at night and see a robot behave differently the following morning. For a student learning to make software meet the physical world, the feedback was persuasive. The experiments also happened among people who would become collaborators, giving the future business something more durable than an idea scribbled on a competition entry.

Before bank documents, there were biscuits

The early computer-vision work had a pleasingly unglamorous range. In 2014, HyperVerge was pursuing railway inspection and an inexpensive system for checking ITC’s biscuit production line. The proposed biscuit system could inspect 2,000 biscuits a minute. There were also licence-plate recognition and photo-organisation products looking for customers. An algorithm’s first audience need not be fashionable to be useful.

That variety matters. Kulkarni’s route into entrepreneurship passed through practical questions about what a camera and software could inspect more cheaply. The work had to fit an industry’s economics. A sophisticated inspection machine is a poor bargain if its cost makes the biscuit beneath it prohibitively expensive. There is a respectable business education in a packet of biscuits.

By August 2015, HyperVerge had announced a $1 million seed round from NEA, Milliways Ventures and Naya Ventures. Its consumer application, Silver, would organise photographs, remove duplicates and poor images, and help people share their favourites. The intended convenience was easy to appreciate: a phone might contain thousands of memories and very little order.

The photograph business meets its own reality

The consumer market was less obliging than the engineering. Silver and Magic Share did not produce the habitual use the team needed. Acquisition conversations emerged as another possible route. In a business built around recognition, there was an uncomfortable recognition to make about the business itself.

Kulkarni’s recollection of this period is unusually candid. The team stopped work for a week to talk. Mentor and investor Venky Harinarayan helped clarify the commercial problem. Sridhar Vembu of Zoho supplied both an example and a customer. The immediate task was to live on revenue; Zoho became HyperVerge’s first paying customer.

“Whose life are we trying to change here and how?”

Kedar Kulkarni

The question is short enough to survive a meeting without becoming a presentation. It also makes an awkward demand of its author. Once you ask it, activity alone becomes a less satisfactory answer. More features, more meetings and more photographs cannot settle whether the work is moving towards something people will pay for and use.

The enterprise turn gave the existing technology a different job. In his account of the recovery, Kulkarni even ended with a joke about timely, recurring payments from clients. Purpose had acquired an invoice. For anyone who has tried to finance a generous intention, this is a fairly consequential development.

Getting through the digital front door

Identity verification made the application of computer vision concrete. A financial institution needs to establish who is opening an account or applying for a service. A customer wants the exercise finished without an expedition through forms and queues. Kulkarni’s business sits between those needs, where speed is welcome and a convincing impostor is decidedly less so.

By 2022, he was describing work with Home Credit that began in its digital channel and extended into offline processes. The collaboration involved teams exchanging ideas and working on the next set of improvements. That detail is revealing: the product’s usefulness depended on understanding how another organisation actually worked, down to the tasks its staff carried out.

In February 2023, Kulkarni described HyperVerge’s product vision as making digital transactions faster, easier and cheaper. He said large enterprises had begun using its AI engines for identity verification by 2018, and that it had onboarded 700 million users over the preceding three years. The customers included telecom operators, lenders and platforms whose daily business required establishing trust at a distance.

Such work is rarely the part of a digital service anyone remembers fondly. Few people open an account for the pleasure of submitting identification. Making that passage less cumbersome is therefore a peculiar kind of achievement: when the work goes well, the person using it can get on with something else.

Two engines, and somewhere to go

Kulkarni describes his preferred structure as a conscious business. In his own formulation, an economic engine earns and compounds profits over time. A contribution engine addresses fundamental problems, including education and employment. Both should last beyond their founders. The commercial side supplies patient capital; the contribution side supplies a reason to keep building.

His example is the Tata group, with operating businesses alongside the Tata Trusts. The appeal is partly a matter of time. A business designed to endure can support work whose results arrive slowly. Education does not conveniently complete its assignment before the next fundraising announcement. Students, employers and families have their own calendars.

HyperVerge Academy gives the idea an organisational address. In 2020, HyperVerge assembled a team dedicated to contribution. The academy began by listening to people and learning from organisations already working on the problems. There is a modest discipline in that sequence: investigate the need before becoming excessively pleased with one’s proposed solution.

The educational thread also reaches back to Lema Labs. Lema Labs went on to train more than 15,000 students in robotics. Seen across the years, teaching looks less like an accessory to his commercial career than a concern that keeps finding another form.

An AI enthusiast who keeps the underwriter

In September 2026, HyperVerge introduced three AI underwriting agents at Global Fintech Fest in Mumbai. They address document analysis, video-based discussions and background screening. Kulkarni described their role as helping a credit professional review information, comparing the result to a cricket highlights reel. The underwriter still has to understand the match.

For complex business loans, he said human involvement remained necessary. His description included links back to underlying records and audit trails. Those details put a useful limit around the promise: a faster summary must remain open to inspection. A system that makes a decision easier to review has a different obligation from one that merely produces a confident answer.

The shift also brings the old commercial question back into view. If assessing a smaller loan costs too much, the lender has less incentive to offer it. Reducing that expense could change which applications are worth handling. The possibility interests Kulkarni because convenience, operating cost and access meet in the same piece of work.

A company with a longer attention span

There are lighter details around the ambition. Away from work, he spends time with his family, mentors entrepreneurs and takes an interest in Warren Buffett. His language about compounding and patience makes that last enthusiasm unsurprising. Some founders collect inspirational quotations; Kulkarni appears interested in whether an institution can afford to keep its promises after the inspiration has worn off.

His stated goal of helping build 100 conscious businesses remains an aspiration. It is also a useful way to read the career so far: a student research group, an education venture, a difficult consumer experiment, an enterprise company and an academy. The settings change. The relationship between useful work, revenue and opportunity keeps returning.

Back in the pantry, the team’s question was what they would choose to do. The years since have added a more demanding clause: how will the work support itself long enough to matter? It is a question fit for a founder with ambitions beyond his own tenure. It also has the advantage of being difficult to answer with a new photo filter.