Pinterest's head of engineering outlines platform future; hints about API

Pinterest's engineering team reveals some of the magic behind all of those colorful pinboards.
Written by Rachel King, Contributor

SAN FRANCISCO -- Pinterest might come off to some as just another social network of digital boards teeming with frilly white dresses, gluten-free recipes, and tips for getting super-toned abs.

But it's also a virtual gold mine for brands and marketers from all corners of business given that the site is just brimming with data.

See also: Pinterest updating privacy policy soon; stresses 'Do Not Track' supportPinterest courting businesses with more tools, updated terms of servicePinterest offers more opportunity to retailers than not | Pinterest named as one of top social trends sites in 2011: report

During a whiteboard session at the startup's new 3-story headquarters in San Francisco's SOMA district on Wednesday, the Pinterest engineering team explained where the growing platform is going next with all of that information.

Jon Jenkins, head of engineering at Pinterest, who was at Amazon for nearly a decade prior to joining Pinterest last September, broke down the recent history of the social network's infrastructure in a nutshell:

  • 2010 was the year of the creation of the business. There was one engineer and two co-founders.
  • 2011 was about scaling challenges given that business was doubling every month.
  • 2012 was the year of mobile. In August, Pinterest released new versions for both iOS and Android.

This year, Jenkins said the focus is now on adding value to pins and extended functionality on top of that.

For instance, Jenkins reflected that the entire site, written in Python for the most part, has been nearly entirely rewritten during the last year.

"It's a fundamental belief of ours that the interests of content providers and our users are aligned," Jenkins declared.

Furthermore, Pinterest runs on a network that somewhat parallels ideas behind the Open and Social Graph infrastructures at Facebook with its own "Interest Graph."

Yet Jenkins also argued that he doesn't think there is a single other company building an infrastructure similar to the Interest Graph.

As far as how the data becomes generated, Jenkins stressed that "a pin can't exist unless it's assigned to a board" as the concept of board "is a really fundamental thing" that carries context.

He continued that part of the puzzle is identifying and solving "adjacent interests." Jenkins described this as having different users pinning similar objects for different reasons.

That's an important aspect of how Pinterest works, Jenkins reiterated, because then the team can learn how to provide content for discovery.

Looking forward, Pinterest engineers are building upon a feature that just debuted last Friday: a new feed editor, which uses the information provided by boards to make recommendations for other content on Pinterest.

Coming up in the next week or two is the contextual menu, designed to present new navigational paradigms among boards. (If you've already seen it, Jenkins quipped that the feature was quietly rolled out to a small beta group.)

Jenkins also revealed that Pinterest is working with a "close set of partners" in brainstorming what an API should look like, stipulating that the team doesn't want to making mistakes, release features, and pull back on functionality later.

"It's a fundamental belief of ours that the interests of content providers and our users are aligned," Jenkins declared.

However, Jenkins declined to offer a timeline or even an estimated drop date for an API.

Still, defining a roadmap could prove to be difficult as more users sign on and Pinterest popularity grows worldwide. Based on Jenkins' comments, it is evident that scaling issues were not limited to 2011.

Jenkins admitted that "figuring out how we're going to scale data repositories" is going to be an ongoing (not to mention complicated) issue.

Editorial standards