TECH TITANS: Growing the Pinterest Team Remotely with Jeremy King, Senior Vice President & Head of Engineering at Pinterest

(Intro Narrator at 00:00:00) For the next four episodes, we're doing something very special. We're taking the best leadership episodes from the most popular past guests and distilling them down to ten-minute segments to turbocharge your leadership. If you want even more ten-minute condensed leadership episodes, you can subscribe to Joel Beasley, Tech Titans in your podcast app. Jeremy from Pinterest joins us in this episode to share his best advice on restructuring your experimentation process and much more. You're listening to Joel Beasley,

(Jeremy (from Pinterest) at 00:00:32) Tech Titans.

(Joel Beasley at 00:00:35) What are you learning right now as a leader?

(Jeremy (from Pinterest) at 00:00:37) Oh, I'm learning how to run a large team when we never see each other. Right? You know, from my time at Walmart, one of the things I did, I spent a ton of time with the team all over the world. And the pandemic's been tough. I mean, trying to run a large organization, you really have to triple down on your communications.

(Jeremy (from Pinterest) at 00:00:57) You have to really understand documentation and build a culture of investing in docs. You know, I was just talking to a woman who was a lawyer, and she was talking about young lawyers and how do they learn law. And, you know, in a law practice, when you're young, you're usually working fifteen hours and you're working with clients and you're working with twenty different clients and you're randomly going to meetings to hear things. And this is how engineering is exactly. And so I've been very nervous about how do we bring on new engineers and ramp them at the same pace as we did pre-pandemic.

(Jeremy (from Pinterest) at 00:01:35) And I know the freedom that the work-from-home lifestyle has given people, especially for folks like you who have young kids and that sort of thing. It's been freeing and amazing. But I look back at my time when I had young kids your age—I was spending, I'd sleep in the office sometimes. And I look back on that part of my career, not only the best time I had in my career from a people and accomplishment standpoint, but I learned more than I ever had. Right?

(Jeremy (from Pinterest) at 00:02:00) I'm just worried that people have forgotten that this is an important part of culture. Right? And how do you build that in an environment where you're completely remote? So that's the hardest thing to do in leadership right now, is to figure out how to balance the fully remote environment. Right?

(Jeremy (from Pinterest) at 00:02:18) We're not quite fully remote. We're coming in once every other week or once a month. Most teams come in about once a month, so people get to know each other, but it's still pretty different. Right? You know, if you can't lean over to your buddy and ask, "Hey, what's that thing again that rebuilds the system when it crashes?"

(Jeremy (from Pinterest) at 00:02:34) You know, it's like, I say everything that can't be solved in Slack that isn't a half-hour meeting is this no man's land. Like, the death of the five-minute meeting. I just need somebody for five minutes. You have to schedule it next Thursday for thirty minutes. Right?

(Jeremy (from Pinterest) at 00:02:50) You know, it's like, and so productivity—I've been worried about productivity. I've been worried about onboarding. And then just worried about, you know, how fast people grow in their careers in this situation.

(Joel Beasley at 00:03:03) I got to talk a few months ago to the CTO of GoDaddy. And

(Jeremy (from Pinterest) at 00:03:08) Oh, great.

(Joel Beasley at 00:03:09) One of the things I was super surprised about is the culture they have for experimentation. It's deeply embedded into their culture. They've built their own systems to, you know, how they can propose experiments, and then you run them and you watch them, and it's one central system that will allow experiments to happen across their entire platform, all these different products that they build and own. And so that's baked into Pinterest's core culture too. Right?

(Jeremy (from Pinterest) at 00:03:34) Absolutely. Yeah. From the beginning. As a matter of fact, when I came here, one of the problems I had with it was—and this will sound very demeaning—but it was our product strategy was often, "We'll run these 100 experiments and see which one works." And I'm like, that's not a product strategy.

(Jeremy (from Pinterest) at 00:03:50) That's an experimentation strategy. So it's great to have it, but you need to be going somewhere. Like, "Hey. We're going this direction, and then here's the 100 experiments that are gonna get us there." But, yeah, it's very common to be running—I mean, we're running—we just upgraded our machine learning systems as well, and we can run a thousand models now instead of a hundred, just to test which ones are, you know, which signals are better for relevance. And so we're just constantly improving, improving that. So both on the machine learning side and on the client side, we're running experiments like crazy. And it is, you know, we don't even think about it.

(Jeremy (from Pinterest) at 00:04:31) The other stat that I use a lot, which I was blown away by when I joined Pinterest, was we have an analytics system that consumes both our big data platform and some of our SQL data. Sixty percent of the company uses that on a monthly basis, like, six-zero. Like, crazy. Like, it's amazing how data-oriented people are in the company about decision making. In my first few meetings when I joined Pinterest, there's this guy, Dan Lohrey.

(Jeremy (from Pinterest) at 00:05:03) He's one of our best product managers. Somebody asked a question in a meeting. Literally, he opened his laptop, did a query, and gave the answer back. And, you know, that kind of thing, having real-time analytics and having a company that answers questions with data is—it's so awesome to work at that, you know, versus the HIPPO, you know, "I just—this is the way we're going. I feel like we should go this way."

(Jeremy (from Pinterest) at 00:05:29) It's literally a data-oriented company. And it's hard to understand how different it was from—for a company who just doesn't have access to the analytics they need in order to get answers to data in real time. And it's amazing to work for a company like that.

(Joel Beasley at 00:05:45) What type of hard problems are you solving over at Pinterest?

(Jeremy (from Pinterest) at 00:05:49) You know, Pinterest is a place where 400 million monthly active people come. You know, they come to do everything from finding dog houses to safe rooms to redecorating their kitchen to "What are they gonna wear for New Year's?" "How do I redesign my look?" you know, from a hairstyle standpoint, "How to find a good pair of shoes," that kind of thing. So it's such a variety of things that people are getting inspired by. We have a ton of people that are gardeners, and they want to know how to do succulents and, you know, design garden space. So it's a big part of the variety.

(Jeremy (from Pinterest) at 00:06:24) And so that means ingesting billions of pins coming through into Pinterest and then using both computer vision and machine learning to do what is content understanding. Like, "What is this pin?" Right? Is it safe? You know, this is obviously one of the most important things about Pinterest is making it the safest place to go on the Internet.

(Jeremy (from Pinterest) at 00:06:45) You know, you don't have any politics. You don't have any suspect content. You don't have any of the crazy stuff you see on the social platforms. So, so number one, making it safe, but then making sure that we really identify, you know, what it is. Some people come here for inspiration, but they're also wanting to learn. Right?

(Jeremy (from Pinterest) at 00:07:03) They want to—we call it sometimes "edutainment." You know, people are, you know, building websites that are teaching them how to cook or, you know, how to garden or that sort of thing. So, and then, of course, all the big hard problems when somebody does a search for, you know, "New Year's Eve gown ideas" or "New Year's, you know, Saint Patty's Day party ideas," you know, I gotta collect a billion images and pins, and then how do I reply back in a good way? So all the hard machine learning problems and the computer vision problems are great.

(Jeremy (from Pinterest) at 00:07:35) But, really, the reason I joined is because 90% of the people that come to Pinterest, they say, you know, "I wish it was easier to find the thing. Once I get inspired, I want to buy it." And I don't know if you saw this, whether it was shelving or, you know, a safe for your room or something like that. Like, where can I buy that safe, right, or that particular item? And it's really about merging all the work that I've done for the last twenty years, you know, of Internet retail, you know, getting connected with the inspiration platforms.

(Jeremy (from Pinterest) at 00:08:06) And so that's really what I've been working on.