Kedar Kulkarni joined IIT Madras as a Metallurgy and Materials Sciences pupil – a department that, as campus lore goes, usually helps you uncover what you don’t wish to do in life. For Kulkarni, that proved true.
What fascinated Kulkarni on campus have been the tech college students on the Middle for Innovation, the place college students burned past the midnight oil to finish initiatives and innovate.
Drawn by their power, Kulkarni joined a robotics staff representing IIT Madras on the Worldwide Aerial Robotics Competitors. As a first-year pupil, he was the least skilled as in comparison with his teammates. Then got here a chance: each autonomous robotic wants imaginative and prescient to navigate – figuring out rooms, studying signboards, and understanding its environment. This required constructing algorithms to assist the robotic “see.” Since laptop imaginative and prescient was largely unexplored inside the staff on the time, Kulkarni’s newcomer standing labored in his favor, getting the liberty to experiment. That hands-on expertise grew to become his first actual brush with robotics—and in the end, his entry level into AI.
“The kick of robotics, or AI, was that you simply spent an evening engaged on it, with simply your laptop, and the following morning, that robotic might be working. That prompt gratification was one thing that drew me to the sector,” Kulkarni says.
The delivery of HyperVerge
Kulkarni’s journey with startups started unintentionally. Two of his shut buddies—now HyperVerge co-founders—had landed prestigious internships: one at Microsoft Redmond, the opposite at Schlumberger. In the future, throughout his internship, co-founder Vignesh Krishnakumar known as Kulkarni with the thought of beginning an organization. Kulkarni brushed it off, urging him to give attention to the internship and revisit the thought later.
When his buddies returned to India, they visited Kulkarni. He laid out the truth: no enterprise capital buzz, no large salaries. He even tried to scare them off—twice. However on the third try, they returned decided to begin.
HyperVerge was born out of a shared need to resolve real-world issues. It was additionally the identical time when deep studying and convolutional neural networks have been taking off. The staff began by constructing an app to type photographs, making strides in face and scene recognition.
To fund the thought, they launched a Kickstarter marketing campaign aiming for $50,000 and flew to the US, hoping alumni may contribute. Their expectations have been modest – possibly a $10,000 verify from somebody. As a substitute, in simply 15 days, they raised an astonishing $700,000.
Rebirth: Constructing a aware enterprise
In the future, the 11-member HyperVerge staff sat in a small pantry, asking: If cash and societal constraints didn’t exist, what work would we wish to do? Solutions poured in—healthcare, schooling, rural employment. Everybody wished to resolve basic issues that actually change lives.
The truth? They’d solely six months of runway left and no clear path ahead.
That’s after they met Sridhar Vembu of Zoho. His story confirmed them that constructing a financially impartial, worthwhile firm allows long-term affect. Zoho’s skill to empower tens of millions of companies whereas contributing meaningfully to rural India deeply impressed them.
That day, HyperVerge was reborn—with a imaginative and prescient to construct a aware enterprise: creating robust financial engines that will outlive them, like Tata or Berkshire Hathaway, whereas concurrently powering contribution engines that drive social change for generations to return. One in all these engines is the HyperVerge Academy, a non revenue academic establishment, devoted to driving social affect by providing free AI and know-how coaching to underserved college students. Up to now, the academy has enabled over 190 college students to safe tech roles, with a median beginning wage of ₹4.2 Lakhs Per Annum (LPA).
The challenges
At this time, HyperVerge supplies AI-powered digital onboarding and fraud prevention solutionsfor sectors like fintech, banking, insurance coverage, and logistics. Up to now, HyperVerge has enabled seamless, compliant onboarding for over a billion customers for 480+ prospects
Early on, the largest problem in constructing AI engines was the supply of coaching information. Reaching near-perfect accuracy—99.5% on datasets of 10 or 100 million data—required tens of millions of photographs. However sourcing actual information, particularly delicate paperwork like PAN playing cards, wasn’t possible. Errors at scale might be catastrophic; even a ten% error fee may spell tens of millions of failed transactions.
The answer? Artificial information era. HyperVerge constructed datasets from scratch, simulating real-world paperwork, errors, and even fraud patterns. This allowed them to coach AI fashions tailor-made for Indian and Southeast Asian circumstances with out compromising consumer privateness.
Fraud stays an ever-evolving problem within the digital world. Unhealthy-faith actors in each society are consistently exploiting loopholes to trigger monetary or operational losses. HyperVerge’s AI engines are constructed to detect and stop these threats—however as fraudsters evolve, so should the defenses.
“It’s a relentless cycle: as we strengthen defenses, new fraud patterns emerge. Over time, this cat-and-mouse sport has change into one thing we embrace—pushing our know-how to remain forward, guaranteeing companies stay safe in an setting the place fraud isn’t just a threat however an inevitability,” says Kulkarni.
AWS and the ability of scale
Kulkarni’s journey with AWS started again in school. Round 2012-2013, constructing deep studying engines was nonetheless confined to desktops and laptops. When AWS launched a program providing $1,000 in credit to pupil researchers, the staff seized the chance to arrange their first cloud occasion, and every part modified.
For the primary time, they may course of tens of millions of photographs in a single day—one thing unattainable for people to attain. This wasn’t simply pace; it was scale and accuracy mixed. Whether or not dealing with 10,000 or 100 million photographs, the cloud made it doable. That have revealed what know-how may do at scale.
Once they began HyperVerge, AWS prolonged $100K in credit—essential for a bootstrapped firm nonetheless chasing product-market match. These credit stretched their runway and stored them going.
At this time, AWS is a core accomplice powering HyperVerge’s mission-critical methods. The corporate works with 75+ banks and NBFCs, the place uptime, safety, and reliability aren’t non-compulsory—they’re every part. If methods cease, onboarding for main monetary establishments halts. AWS allows them to take care of the best requirements of availability and safety, with robust safeguards towards fraud and assaults.
Earlier, the large debate for delicate purposes was all the time on-prem vs. cloud. At this time, AWS has earned that belief. Mission-critical purposes run seamlessly on its infrastructure, permitting startups like HyperVerge to scale effortlessly. Establishing an in-house information middle would have required months of procurement, permissions, and setup. With AWS, scaling from 1 lakh prospects a month to 10x and even 20x that quantity is frictionless.
“AWS didn’t simply give us infrastructure; it gave us agility, resilience, and confidence. For corporations like HyperVerge, that has made all of the distinction,” says Kulkarni.
AI, ethics, and accountability
AI is advancing at a wide ranging tempo, with each group racing to construct the following breakthrough mannequin. However this pace brings challenges—moral issues, workforce readiness, and security. One main space is jobs: as new roles emerge, will individuals have time to reskill?
Equally important is decision-making. In sectors like healthcare and protection, we can’t enable methods to take ultimate calls. AI can help, however a human physician should carry out surgical procedure; a human should approve important protection choices.
Then there’s bias. Human prejudices usually seep into fashions, making equity an ongoing problem for AI security. Stopping these biases whereas constructing methods at breakneck pace might outline the following decade.
The street forward: AGI and Bodily AI?
Kulkarni famous that present AI instruments are spectacular however nonetheless closely reliant on human judgment. Whether or not producing content material or code, outcomes may be hit and miss. True AGI (Synthetic Normal Intelligence), the place human oversight is minimal, stays the large query however nonetheless excites Kulkarni. Many predict the arrival of AGI by 2025–2030, however the path and challenges forward are unclear.
One other fascinating frontier is bodily AI. Whereas in the present day’s breakthroughs are confined to digital duties like writing or coding, replicating human dexterity within the bodily world is much extra complicated. From making a espresso to assembling a tool, the hole stays big. Closing it would demand advances in physics simulators and strong methods for real-world interplay, requiring huge innovation and assets, however one that might yield nice innovation.
Cultivating joyfulness
Outdoors of labor, yoga has been a gentle a part of Kulkarni’s life for over a decade. He believes that to construct a aware enterprise that creates pleasure, one should first be joyful themselves—and yoga helps him do precisely that.
He additionally spends most of his free time together with his three-year-old daughter. Watching her develop, study new issues, and develop intelligence is each humbling and interesting—a every day reminder of how outstanding human potential actually is.
