Desi tech solutions thrive at Amazon

BENGALURU: Seattle-based etailing giant Amazon has invested heavily in technology platforms such as artificial intelligence, data analytics and machine language to increase customer footprint and reduce seller pain points, a senior executive told ET.

To boot, with a strong team of data research scientists with capabilities in machine and deep learning, computer vision, natural language processing to perform analytics across data pipelines and processing infrastructure, India continues to be Amazon's largest development centre outside the US.

And not just in size, Amazon India CTO Dale Vaz said all products and services have been built locally by Indian engineers, with the nascent Indian ecommerce space throwing up lot of innovation potential. He said poor internet connectivity across India despite a growing mobile internet traffic was the main reason to develop Micron, a tech which helps build light mobile apps to suit low-end smartphones.

"We introduced near-automated sign-ups for first-time mobile users, revamped our daily deals offering and removed elements which were too heavy for the low-end users. The app which was 17 MB in size was reduced to 2 MB." He said over 80% of their traffic comes from mobile devices and their app is being customised to fit different mobile platforms and device screen sizes. The app also detects the customer's location to show widgets offers depending on the network strength and location.

Considering the large amount of data that Amazon churns out every second in India, Vaz's team has built the India Data Mart which contains data sets on customers, their preferences, selling history, seller's details, and market trends. They use these millions of data sets to target customers with right products and help sellers push ads on items through details on customer purchases. "Based on the data, we help sellers who have lesser experience in online advertising, in putting up ads on se lect items. Similarly, our systems analyse customer buying data and pitch items to them which they might need in future."

The data analytics platform also provides insights back to the sellers based on the buying patterns of customers. Vaz said they also use an AI platform to figure out vendors selling overpriced items, items with incorrect details and even pick out fake reviews on sites. The AI model studies data patterns and picks out the ambiguity in terms of user behaviour on the site, a listing of products and time taken by a user on the site.

"Finding authority for an item is where AI comes in. Historically, a person has been doing this but with data splurge, the accuracy in picking out flaws becomes paramount." He explained that a genuine customer visits the Amazon site multiple times, views more than four products, looks at product details and then makes a purchase.