With AI and machine learning, Flipkart reinventing its backend technology

In a bid to ensure that its back-end technologies are able to scale along with its fast-growing online retail business, homegrown e-commerce major Flipkart has embarked on a programe called “Continuous Platformisation”. The programmme is aimed at giving its technologists the bandwidth to build products that can support unprecedented scale whenever it is required rather than just looking at the current requirements.

While shoppers on Flipkart’s website and mobile apps may not notice any difference, the programme will bring in more stability to the platform, reduce the number of bugs and enable faster roll-out of newer features, which will in turn boost a customer's experience, said Utkarsh B, principal architect who is driving the programme at the Bengaluru-headquartered e-commerce company.

Started in 2016, the plan to revamp all back-end platforms emanated from issues Flipkart faced during its Big Billion Day sales. With the platform crashing due to high load in the first year, by the second year the company had made provisions to handle the load, but they were just enough.

“We technologists are always fighting a rat race to always catch up with the business growth. We don’t get the time to pause and rethink that if we had more time, more space, more comfort, then can the state of systems be a lot different than where they are today,” said Utkarsh, who was then asked by Binny Bansal, who had just become the CEO at Flipkart, to steer the programme.

For Flipkart, the approach of continuous platformisation represented a massive mindset shift. After brainstorming about what the ideal product from each team should look like, keeping in mind the massive scale that the company would require, the cogs of change were set into motion.

“It’s a continuous process, I don’t think there can ever be an end to it,” said Utkarsh. “Everything starts small but you never know when these things take gigantic proportions. So you cannot reshape the technological systems in retrospect, you need to continuously transform them.”

The flipside of carrying out the exercise was that teams became more visionary when it came to blueprinting a new product or feature. With Continuous Platformisation set into action, the teams at Flipkart are no longer looking at just the current requirements to come up with product specifications. Knowing that any solution they build will be out-sized in a matter of years if not months, technology is now built keeping scale as an infinite variable.

With Continuous Platformisation, Flipkart is looking at implementing technologies like artificial intelligence (AI) and machine learning (ML) more easily as teams are already rethinking products constantly. Sachin Bansal, co-founder and executive chairman, is leading an internal unit dubbed “AI forIndia” which will look to infuse AI and ML at the core of its business.

While there will be a key set of leaders looking at the push for AI, no separate team is being created at Flipkart that will do this, said CEO Kalyan Krishnamurthy. All business units and technology teams, including internal divisions, are being mandated to figure out where they can implement AI and ML and then execute the same.

“AI is not a project that you hire one person from somewhere and say this person is going to enable AI for Flipkart. The company, which will be actually successful in using AI, will be one which embraces it at every functional level, and it should not be like a special project,” said Krishnamurthy. “This is going to be in the culture and the DNA of Flipkart.”