Interview Careers

Hyperverge: An early start and growing

The Hyperverge team. Photo: Special Arrangement  

A group of five alumni of IIT-Madras — Kedar Kulkarni, Vignesh Krishnakumar, Kishore Natarajan, Saivenkatesh Ashokkumar and Praveen Kumar — started Hyperverge, a company to develop deep-learning software, while nearing the end of their B.Tech. programme. Starting with a seed funding of $15,000, the group has now raised a million dollars for the startup. In this interview Kedar Kulkarni, CEO of Hyperverge, shares his experience.

What does Hyperverge do and what is the kind of development that's gone into the software?

HyperVerge is a deep-learning startup based out of California with its India office in Chennai.

The company has developed patent-pending technology for organisation of images by identification of people, places, scenes and events in images. The basis of the technology lies in the field of machine learning. The team uses millions of images to build/train convolutional neural networks. These are computer programs that can look into an image and find out important details about the image. This is like teaching computers to mimic the human behaviour of interpreting the images that we see, just the way the human brain does it.

Silver, the first smartphone application developed by the company, uses the technology to help users automatically organise their photos, eliminate poor quality photos and duplicate and share their best photos. HyperVerge aims to solve the everyday problem of photo organisation for a large number of smartphone users.

Development efforts include developing new mathematical models, training them with large number of images, testing their accuracies and improving the models. HyperVerge has been able build sophisticated technology models for different functions such as face recognition, scene recognition, image quality. Each model requires millions of images for training and testing.

What are the possible applications of this technology?

The current application that HyperVerge is focusing on has a great scope of enabling many applications that will be deeply integrated in the life of users. These include shopping by pointing your phone, finding out healthiness of your food, and so on.

These applications will become more prominent with the launch of wearables as a digital eye will always be available to capture live images and give back the information to the users.

You have recently raised a million U.S. dollars venture capital for Hyperverge. How did this come through?

Our first round of funding (approximately $15,000) was given by IIT-Madras Incubation Cell and Lema Labs (Kedar’s first company). All the founders of HyperVerge met each other at the Centre for Innovation, IIT-Madras. IIT professors and the ecosystem have mentored and encouraged us through many international competitions, consulting assignments, and so on. Hence, the founders reached out to the IIT-Madras Incubation Cell for initial funding when they decided to start HyperVerge.

The team worked for four months on prototyping the technology. Their first idea about the next round of funding was to raise about $50,000, through a kickstarter campaign in US. Kickstarter is a crowdfunding platform. However, after the mentoring sessions with alumni entrepreneurs in the U.S. and visits to universities such as MIT, CMU, Stanford, etc., the team decided to raise money through venture capital firms in the U.S. This was strategically very important as the space was emerging and large companies such as Google, Facebook, etc. had already started mobilising resources and hiring the best researchers in this field.

Raising money was a team effort. Vignesh and I spent one month in the U.S. meeting investors while Kishore and Praveen managed the India team and supported them.

Eventually more than seven VCs wanted to invest in HyperVerge. The team chose NEA (NEA is the largest VC fund in the world with an amazing tech portfolio), Milliways Ventures (because of their deep understanding of technology and consumer space) and Naya Ventures (because of their experience in mobile, cloud and big-data space). The team also invited Srikanth Sundararajan (Venture Partner at Helion) to invest.

In the early stages the team was mentored by IIT professors. The team received tremendous support from the IIT-Madras Alumni community during and post the fundraising process.

How many people are employed now?

The HyperVerge team comprises 21 people. The India team has 12 full-time team members and eight interns and the U.S. team comprises three consultants.

Who was the person who started it all? Which part came first — the idea of entrepreneurship or the idea of image recognition and deep learning?

Interest and passion for technology of computer vision and machine learning came much before entrepreneurship. The team was working on computer vision for four years before the birth of HyperVerge. Earlier, their interest was purely academic and they participated in many international competitions through 2009-2014.

In 2012, the team wanted to explore real world applications of the technology and they started taking up consulting assignments. They worked with Indian Railways, ITC, MRF and many other companies to develop pilots projects. In 2013, the team decided to take up entrepreneurship and started HyperVerge Technologies.

You are all alumni of IIT-Madras from different batches. How did you come together?

We all met at the Computer Vision Group at IIT-Madras. I joined IIT-Madras in 2008 and founded the Computer Vision Group (CVG) in 2009. CVG is a student research group at IIT-Madras focused on the technology of Computer Vision. Every year, 30 students are selected among 100-130 applicants and are trained on this technology. At the end of the process, 4-5 students are selected for projects that the group undertakes.

Sai Venkatesh, who joined IIT-Madras in 2009, was my junior. He co-led the team in 2010-2011. Vignesh and Kishore joined IIT-Madras in 2010. They co-led the team in 2011-2012. As part of this group, we took on many academic projects, industry consulting assignments and participated in various robotics competitions. We developed great rapport with each other during this time. We were looking for a powerful real world application of our technology and the idea of HyperVerge happened.

This article is closed for comments.
Please Email the Editor

Printable version | Jan 18, 2021 12:51:40 PM |

Next Story