Every summer Pinterest welcomes a new class of software engineering interns to our San Francisco HQ. Each intern (or Pintern, as we say) works on an important technical project that directly impacts our product. Pinterns demo their projects weekly in front of the whole company at Friday Q&A, hosted by our co-founders and other leaders across Pinterest. In this post you’ll learn about some of this year’s class and what they built during their internship.
Caroline Lo | Stanford University, PhD ’17
Pintern Summer 2015, 2016, 2017
I’m Caroline, a PhD student studying computer science at Stanford. My research focuses on understanding human behavior and online content, which made interning on Pinterest’s data science team this summer a perfect fit! The purpose of my project was to analyze and better understand how Pinners use Pinterest’s in-app browser on their phones. This is important because any time a Pin is clicked a user is taken to the website in the in-app browser. This was an open-ended project and we weren’t sure what we’d find or learn. After poking around the dataset, I was able to explore not only general user browsing patterns, but also how the browsing patterns affected other behavior on Pinterest. My project took several interesting detours along the way, but I ultimately focused on analyzing and modeling the shortterm and longterm impact of a website’s quality on in-app browsing and user engagement. There was no established solution or strategy for quantifying this impact, but I experimented and tried all sorts of metrics and models to find the best strategy. For example, it’s not necessarily clear what a “quality” website is, which gave me a great excuse to investigate multiple indicators of website quality and come up with a few myself.
I really enjoyed the flexibility and open-ended nature of my internship project, which enabled me to get creative. Many thanks to Pinterest, the data science team and Brian Karfunkel for a super fun summer!
Dipa Halder | University of California, Berkeley, BS ‘17
Pintern Summer 2016 & 2017
My first experience as a Pintern was last summer 2016 working on the shopping team. I was blown away by how the Pinterest culture is as creative and inspiring as the product itself. As someone who’s passionate about art and computer science, I felt I had found a place that truly valued both. So it was an easy decision to go back and intern with a new team, Search Product and Visual Discovery. I challenged myself with more responsibility and a new set of technical problems. Even though I was new to building product features, I jumped at the opportunity to work on the Lens redesign for Android, my first project of the summer.
Lens for Android, our camera search tool, was ready for a fresh look with several updates including the ability to zoom in and out, tap to focus and prominent camera roll access. One of the bigger pieces I built for its new interface was the carousel at the bottom which allows Pinners to discover Lenses to try or quickly access recent photos. It was challenging to write code that supported various camera hardwares and API levels across different Android devices, but it was a great learning experience that challenged me to find creative workarounds.
While I had some prior experience with Android development and pieces of the Pinterest codebase from previous internships, I ventured outside of that comfort zone to work on platforms I had never touched before, such as the Visual Search API layer and even iOS. The learning curve was steep, but digging into those systems gave me a deeper understanding of how visual search technology is structured at Pinterest, from the way it computes visual features to how those features are translated into results in Lens.
I learned a lot by trying new things this summer and discovered that product engineering is my sweet spot where I get to work closely with design and still write code to build tangible features. I also became a more confident engineer and it was extremely rewarding to go from knowing nothing about Visual Search at Pinterest to seeing my work on Lens launch to our users.
Mira Baliga | University of Maryland, BS ‘20
Engage Pintern Summer 2017
I had no idea what to expect from the next eight weeks of my life as I walked into Pinterest on the first day of my internship. I was an intern with Pinterest Engage, a program for rising college sophomores that introduces them to life at a tech company and provides support needed to grow as engineers. Software engineering was this murky, mysterious concept I knew little about prior to my internship because it wasn’t yet covered much in my computer science classes. In this internship, I got a firsthand look at just how many different specializations it takes to run a tech company, from frontend to backend, to performance and, security–the list goes on and demonstrates how important it is to work cross-functionally. You can’t just be a talented engineer you need to be able to collaborate with other teams, like design and marketing, to build a cohesive product. I learned about career paths like product management, quality assurance, technical writing and other roles that require some technical knowledge, but also creativity, leadership and other skills.
I’m now more confident to continue pursuing a degree in computer science, because I know there’s a technical role out there that’s perfect for my unique talents. Internships are meant to be a time to explore, try new things and ponder some of these questions, so that you can find a role you’ll enjoy when you look for full-time employment. And now, thanks to Pinterest, I’m closer to answering some of these questions myself.
Sen Wang | Carnegie Mellon University, MS ‘17
Pintern Summer 2017
I had a wonderful time working with the discovery team on recommendations. Throughout the summer I mainly worked on two machine learning projects: improving the performance of our “best Pins” models and building better user-independent Pinnability models.
The goal of the best Pin project was to serve Pins with the highest quality to new users to help them discover great content. By applying a linear regression model to the existing data, I built a function that precisely predicts the engagement probability for Pins. This new function leads to a 1 percent increase in weekly Pins saved, which is a huge gain given the size of our user base. The Pinnability model is an important component of Pinterest’s recommendation system. By using a Pinnability score, we can rank Pins and provide Pinners with content that they’re interested in. And, by tuning the old TensorFlow models on user-independent data, our new model increases Pins saved statistics by 1 percent.
With help from my teammates and advisor Daniel Liu, I was proud to produce these great results and positively impact the recommendations engine that serves more than 3 trillion ideas to Pinners every year.
If these types of challenges sound interesting to you, join us!