Lava Kumar and the quiet business of knowing how people feel
He spent nearly two decades shipping products at Yahoo, Motorola and Operative. Then he flew home to Bengaluru and bet his career on a stranger idea: that a machine could read a human reaction, and that reading it well was worth building a company around.
In most product stories, the founder falls for a technology and then spends years searching for a problem to point it at. Lava Kumar ran the tape backwards. Before Entropik built anything worth demoing, he sat down with more than eighty of the largest brands in the world - ITC, Tata, Procter & Gamble among them - and asked a plain question: where does knowing how a customer feels actually change what you decide to do?
That instinct, put the decision first and the machine learning second, runs through everything he has built. Kumar is the co-founder and Chief Product Officer of Entropik, a Bengaluru company working in a corner of artificial intelligence with an unglamorous academic name, affective computing, and a very human promise: software that can read a face, a voice, a gaze and a flicker of attention, and tell you something true about the person behind them.
He did not arrive there by accident. He got there by spending nearly twenty years learning how products are actually made.
The long apprenticeshipTwo decades before the leap
Kumar trained as an engineer at the University of Madras, then went on to an MBA at the College of William & Mary in Virginia. It is an unusual pairing, and it shows in how he works. He can talk to the people writing the model and the people signing the cheque, and he tends to translate between them.
The resume that followed reads like a tour of the adtech and martech industry as it grew up: product roles at Wipro and Motorola, then Yahoo, where he worked on the Gemini advertising platform, then Operative in New York, then Integral Ad Science. These are not household names outside the industry, but inside it they are the plumbing of how digital advertising gets bought, sold and measured. Kumar spent those years shipping consumer and enterprise platforms across markets, and quietly collecting a conviction.
The question sounds simple. Answering it honestly turned out to require a company.
The homecomingLeaving the safe version
In 2016 Kumar left his US martech career and returned to India to co-found Entropik with Ranjan Kumar and Bharat Singh Shekhawat. On paper it was the harder path. Emotion AI was not a category investors were competing to fund. There was no obvious playbook, no wave to ride. What there was, in Kumar's telling, was a gap between what people say in a survey and what they actually feel while they experience something.
The founding team split the work along its natural seams. Ranjan Kumar took the chief executive seat. Bharat Singh Shekhawat, who had spent fifteen years designing data-intensive systems, took engineering. Lava Kumar took product, which in a deep-tech company means the hardest translation job of all: turning a research capability into something a marketing director will pay for and use on a Tuesday.
The disciplineROI before the algorithm
This is where the eighty conversations matter. Plenty of Emotion AI demos can make a chart of your feelings. Far fewer can tell you what to do with it. Kumar built Entropik's product, called Decode, around use cases the brands themselves had told him were worth solving. The measurement had to earn its keep.
It is a striking thing for a product leader in an AI company to say out loud. The temptation in this field is to sell the wonder of the technology. Kumar keeps pulling the conversation back to the decision it changes. That discipline is probably why the platform stuck.
The four signals Decode reads
Illustrative weighting of Decode's multimodal inputs, not exact model coefficients.
By the numbersWhat the bet has produced
In February 2023 Entropik raised a $25 million Series B led by Bessemer Venture Partners and SIG Venture Capital, taking its total funding to roughly $35 million. Money is a lagging indicator, but it is a useful one here: it means a category that barely existed when Kumar came home now has serious institutional backing. Under his product leadership Entropik rolled out Decode 2.0, built to deliver faster insights at enterprise scale, and pushed multimodal analysis across all four signals at once.
The arcHow it came together
What comes nextEmotion as a plug-in
Kumar's ambition is not to keep Emotion AI inside a market-research tool. He talks about exposing it the way generative AI got exposed, through public SDKs and APIs that any developer can wire into their own product. The comparison he reaches for is OpenAI: take a capability that used to require a specialist lab and make it a line of code. He has spoken about extending the same emotion-sensing approach beyond advertising into fields like healthcare and education.
Underneath the roadmap is a belief he states more plainly than most technologists would dare.
It is both the pitch and the responsibility. A tool that reads emotion is powerful precisely because emotion is intimate, and Kumar frames the work as building a transparent, honest layer rather than a surveillance one. Whether the industry holds itself to that standard is a bigger question than any single founder. But it is telling that the person who has spent longest building this particular version of it keeps returning to the word personal, and keeps insisting the measurement is only worth anything if it helps.
There is a tidy symmetry to the career. He started by learning how advertising decides what to put in front of you. He ended up building the thing that might one day tell an advertiser, or a teacher, or a clinician, how you actually reacted. The engineer who became an MBA who became a product leader has spent his whole working life circling the same subject from different sides. The subject is people, and how badly we understand each other, and whether a machine can help close the gap without making it worse.