Check out the on-demand sessions from the Low-Code/No-Code Summit to learn how to successfully innovate and achieve efficiency by upskilling and scaling citizen developers. Watch now.
While eBay could have used any number of existing AI platforms to enhance its various products, the company instead elected to build its own AI system — dubbed Krylov — in-house and make it open source for anyone to use. That decision appears to be paying off.
The San Jose-based company has made no secret of its AI ambitions over the past four years, hoovering up technical talent via acquisitions and launching myriad automated tools to spot credit card fraud, improve product listings, and bring other nifty features to buyers and sellers. Late last year, eBay also relaunched a standalone vehicle-focused Motors app, with AI and automation at its core.
Such is the pervasiveness of AI across eBay in 2020 that the company’s chief technology officer, Mazen Rawashdeh, says AI is now an “ecosystem” at the company.
“I don’t call AI a product — I call it an ecosystem,” Rawashdeh said during a talk at Transform 2020 today. “Because it’s moved into almost everything we serve for our customers, for everything we do internally to manage our platform. We use it for many areas from image search to personalization, recommendations, price guidance, fraud detection, and shipping estimates. So it is absolutely an ecosystem for us.”
Intelligent Security Summit
Learn the critical role of AI & ML in cybersecurity and industry specific case studies on December 8. Register for your free pass today.
These various AI use cases help to illustrate why eBay bothered to build Krylov in the first place: It all comes down to scalability and enabling a “shared model” between teams. According to Rawashdeh, Krylov has helped scale AI across the company and give all the business units easy access to natural language processing (NLP), computer vision, and machine learning smarts, while also bypassing the need to stitch together components from various other AI platforms.
“So eBay had some form of AI capabilities three or four years ago, but it was disjointed, and it was a siloed model,” he explained. “It was not a capability that — end-to-end — takes the process of data, of compute, of shared model, and allows our researchers and data scientists to go to one place and run their model at scale.” Thus, Krylov effectively enables eBay to centralize its AI, saving individual researcher and development teams from running resource-intensive models individually.
The COVID-19 crisis has also shone a light on some of the benefits of operating a centralized end-to-end AI platform. Both Amazon and eBay have been criticized in recent months over so-called “price-gouging” practices, whereby sellers vastly inflate the cost of high-demand items such as face masks and hand-sanitizers. As things transpired, the company’s existing system was already running models of 100 million uploaded product images at at time, and this proved useful for each of eBay’s teams that were looking to automate the process of taking down listings that violated its policies around price-gouging and prohibited items.
“We created a really deep understanding of those images, where object detection, image classification, and textual recognition is available for the rest of the different business units,” Rawashdeh said. “They didn’t need to worry about getting those expensive models — they were already available, and we leveraged them. As a result, we were able to take down over 15 million items on eBay.”
VentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative enterprise technology and transact. Discover our Briefings.