Interview with Biswanath Patel, Co-Founder at Buyt


One of the most difficult tasks for a buyer today is to select a product from the long list of options that are available to him. The number of ecommerce retailers are on a rise and so is the number of companies making a particular product. So with this ever increasing list of options to make the selection from, the need of a product comparison platform is quite natural. While comparison works in standard categories like electronics, discovery and recommendation is required in aevery facet of e-com, be it fashion e-tail, services like insurance and bank products or even travel. is one such product discovery and recommendation platform for online shopping. is an online product discovery and recommendation platform which integrates commerce with content and enables information flow from a merchant or service provider to the consumer. The company which was founded by Yogesh Tomar and Biswanath Patel in the year 2012, and later joined by Abhishek Agarwala in 2014, offers product comparison through a number of popular websites and uses a copyrighted intelligence e-Commerce Metasearch Engine which they have built in-house. This New Delhi based startup is already among the top marketing partners and traffic drivers for leading ecommerce sites such as Flipkart, Amazon and Snapdeal. In July last year, it got acquired by DailyHunt (formerly NewsHunt), India’s largest and mobile-focused local language news platform, for an undisclosed deal amount.

Here is a small interview with Biswanath Patel, the co-founder of the company.

Give us a snapshot of your professional journey till now.

After graduating from IIT in 2009, I joined the Equity Quantitative Analytics Team at Nomura India. I worked on the development and deployment of Quark – a novel, Python and Java-based WAN distributed framework used to crunch large amounts of market data for pre and post-trade analytics, which forms the core of Nomura’s flagship analytics offering, TradeSpex™. During the second half of my stint, I was part of Nomura New York’s Systematic Trading team to build scalable data scraping and processing utilities to collate real-time data from various news sources and feeds, which formed the backbone of very successful intra-day trading strategies. We handled data services libraries which disambiguated, processed and updated raw data from news streams, data vendor feeds and the web to create algorithmic trading strategies in the worldwide equity markets. As a small team of 6, we used to trade on markets from Australia to the US, and managed a portfolio of over $100m with returns of over 30% consistently for 3 years.

By 2012, internal hedge funds and trading was discouraged at big banks, and Nomura decided to spin off its trading units. I moved back to India to start BuyT with my Nomura colleague, Yogesh Tomar and it has been a great journey so far.

When & how did the entrepreneurial bug bite you?

Even during my IIT days, I had gotten interested in start-ups, having heard of early successes like Flipkart and Zomato. While we were trading in stock markets, I came across several entrepreneurial success stories each increasing the “itch” so to say. The biggest was of course Facebook which had its IPO while we were active traders. We made a ton of money off Facebook’s stock, but that’s another matter :). The winding down of trading teams at Nomura provided an apt time to come back to India and plunge headlong into entrepreneurship. E-commerce in India was booming and I had a good experience in big data curation, analysis and extraction of actionable insights to make decisions. For me, it was the perfect opportunity to start working on an e-commerce meta-search and recommendation platform (BuyT).

What were the early days at Buyt like?

It was a typical ‘garage start-up’. Me and my co-founder, Yogesh were the only people involved for the first 6 months (January to June 2013). We were hardcore coders. Our initial investment was our 2 machines, and a couple of cloud servers to host our code and collaborate. We used to work at our home-cum-office (which used to be the hall of our residence). From the very beginning, I realized that the core of any company was the “product”, so in the first few months, we invested all our efforts into building a world-class working product. Our meta-search engine, was ready – it didn’t have all the bells and whistles, but we ensured that our recommendation engine was the best-in-class. Obscure products like cushion covers and mobile accessories could be searched and found on our platform even better than e-commerce sites themselves!

Can you please describe what a typical day at office is like?

BuyT is on a mission to help people make the most informed decisions while availing digital services and transactions. On one hand we collate and curate data from service providers (for example, e-commerce stores), on the other, we crawl and analyze content from major publishers to understand and interpret context and user behavior. Merging these two enables us to recommend the best deals, goods and services right where the content is and when a user is looking for information about them. We are all about performance and ROI driven marketing. Currently the BuyT ContentLink platform has been rolled out with marquee Indian publishers like TOI, IndianExpress and Jagran, serving over 100 million monthly recommendations. There are multiple product and engineering teams working at BuyT and a typical day in office includes a session on brainstorming on innovative ideas to take our algorithms to the next level, taking stock of existing systems, and continuously striving to automate our ongoing processes.

Can you tell us about the technology stack used at Buyt?

Since we’re involved in the entire lifecycle, from data scraping to recommendation and data rendering, we use a diverse tech stack. Most of it is open-source and cheaply available. We believe scalable systems anyway require too much customized programming to be able to use expensive off-the-shelf softwares. Our stack includes Python, PHP, Java, Linux, nginx, MySQL, mongoDB, other NoSQL formats, JavaScript, redis, Solr, numpy/scipy and other ML toolkits.

Is there any technology that you’re personally betting on to help you scale up?

I believe that good technology uses simple things at scale and needs to be developed in-house using existing technologies – fancy off-the-shelf solutions cannot usually be purposely modified to scale up. Instead of pinpointing a single technology, I will say recommendation as a problem has not been successfully cracked yet and has a huge potential. The world’s major tech companies – Google, Facebook, even Amazon are basically very good recommendation platforms in various forms – Google for search queries, Facebook for news and shared articles and Amazon for products. At BuyT, we made a conscious decision not to remain an e-commerce search engine, but to use our core expertise – data analytics on users and their interactions (clicks, views, transactions) to predict the best products and services to them – to grow our platform to a broad-based recommendation engine, be it for articles, products or even ads. Currently we work with all the big content publishers in India – Times of India, NDTV, Indian Express and most importantly, Dailyhunt, India’s largest content aggregator app.

How are you using Business Intelligence and analytics at your company?

In today’s information driven world, data and analytics are the most important, in fact the only defence that a technology company has. So it is of course our most important asset. Through our partnerships with publishers, we currently serve over 100m page views to 40m unique users per month. The impressions served to each of them, their clicks and transactions all feed back into the recommendation system on a real-time basis to continuously enhance the targeting that we do. On the other hand, our catalog of 50m products from across all e-commerce stores gets updated in a matter of hours. Our analytics platform crunches data from all these users to keep us ahead of the game and make our platform the best-in-class.

How else does technology help your company stand apart from its rivals?

Strictly speaking, no one else in India is currently working on a “recommendation” product with the reach that we have. Our huge user base allows us to analyse GBs of data every second and fine-tune our machine learning algorithms to improve recommendations. While our current focus is product recommendation from e-commerce stores, we soon plan to expand the engine to other services, content and even ads on the Dailyhunt platform. Dailyhunt is India’s largest content aggregator app, with over 25m Monthly Active Users (MAUs) and is owned by our parent company, Ver Se Innovations Pvt. Ltd.

What has been the biggest technical challenge you’ve faced while running Buyt?

I think the core tech companies have not really been encouraged in India due to a multitude of reasons – a tradition of “jugaad”, which while helpful prevents a perfectionist view of things that flourishes in Silicon Valley for example, the lack of quality start-ups and companies in the domain, the relentless focus and heavy investments in ideas (because of a 100bn market) and the number of users without actual technical depth, etc. Of course, this is slowly and steadily changing and companies like Flipkart are investing heavily in tech. So our initial challenge was to convince our partners and investors of the potential and “ability to scale” of a recommendation engine. After initial scepticism, we were able to convince all stakeholders and currently lead the market for such a product.

What are some of the attributes you look out for in prospective technology employees?

My belief is that specific technologies are not difficult to master if a person has the right talent and temperament. For example, someone who knows C++ can master another language like Java or Python quite easily as the paradigms of programming are similar across languages and the only thing that varies is syntax. So I look for general programming proficiency, problem solving ability, a basic level of fluency in mathematics, and above all, a “can-do” mentality. For learning the rest, there is Google! 🙂

How do you keep up to date with the latest happenings in the technology world?

Mostly online. I have subscriptions to a lot of tech and start-up-related groups like techcrunch, techcircle and vccircle. The rest comes by voraciously reading articles and tech magazines, and also talking and brainstorming with colleagues and the team. Sometimes, events like start-up and tech meets do help.

What gets you excited about coming to work every day?

I worked on expectation maximization (EM) algorithms and machine learning during my internship at the Max Planck Institute of Biological Cybernetics in Tubingen, Germany while I was in the 3rd year at IIT Bombay. That piqued my interest in machine learning and how a system could, on its own, “auto-learn” based on data and some initial ad-hoc rules. BuyT finally gave me the opportunity to merge this theory into the real-world commercial applications. This being my initial driving principle has continued till today. After almost 3 years, we still keep on innovating and even now, we’re constantly broadening the scope of the application of our recommendation engine – so there are new things to do and more than enough to keep me and the team motivated.

What sets Buyt engineering culture apart?

Our single-minded focus is building good, useful technology and cutting-edge products. We’re flexible in most other ways – whether people want to shift their work hours, or work from home for a while or schedule their leave flexibly. While discipline is of course important, and we do have strict deadlines, I believe a lack of everyday constraints allows people to work at their productive best.

Which Buyt value resonates best with you?

Simplicity is the best policy :). As I said before, instead of fancy-sounding and expensive software, we have predominantly used open-source software which are easy to use, learn and customize. These first principles have helped us scale to 40m unique users and 100m page views per month!

Any piece of advice for the techies out there?

A particular language does not matter. Improve programming skills in general and always be on the lookout to improve. And refer to (14). 🙂


Leave a Reply

Your email address will not be published. Required fields are marked *