2 min

Tags in this article

, , ,

Over the past two years, eBay has been improving its information technology infrastructure. Now the company has reached a milestone. The webshop has finished its own designed servers. At the end of the year the server designs are made open-source.

The plan was to create its own hardware and an artificial intelligence engine, reports Silicon Angle. In addition, his data centers had to be decentralized and the architecture had to be set up for edge computing. All changes should help the company use newer technologies such as Kubernetes, Envoy Proxy, MongoDB and Apache Kafka.

The company has now announced that the servers will be finished and deployed after nine months. According to Mazen Rawashdeh, vice president of platform engineering, it can often take years to reach the same point. However, eBay could do this in “an aggressive timeline”, at no extra cost.

“We have systematically gone through every layer of our technology, studying efficiency, capability and the possibility of improving existing solutions.

AI engine

The company claims that it can already benefit from the new infrastructure. For example, new technologies are being used to build an in-house AI engine that allows developers to create new products and experiences for customers.

“Our AI engine has already accelerated the production of new features such as computer vision, Image Search and social media sharing. This allows teams to reduce the development time from weeks to just hours,” says Rawashdeh.

The Image Search API was introduced last month. This allows developers to use the visual search services that eBay has implemented to help users search the website. For example, a user can take a picture of something they want to buy and have eBay search for similar products.

This news article was automatically translated from Dutch to give Techzine.eu a head start. All news articles after September 1, 2019 are written in native English and NOT translated. All our background stories are written in native English as well. For more information read our launch article.