Create the engineering career you love at Pinterest
An interview with Behnam Rezaei | Pinterest VP, Engineering
At Pinterest, we’re on a mission to bring everyone the inspiration to create a life they love. For our employees, this extends further to creating the life and career they love. The Pinterest Engineering Blog team sat down with Behnam Rezaei to get an inside scoop into the Monetization Engineering team, what makes Pinterest different and why now is a great time to join our team.
Joining Pinterest in March 2023, Behnam Rezaei is Pinterest’s VP for Monetization — Machine Learning Engineering and Data Science based in San Francisco.
Can you share more about your team at Pinterest?
What’s the goal of the team? What are the biggest opportunities you see? What are you most looking forward to?
Pinterest has three main engineering organizations: infrastructure which is an enabler for various teams, core engineering is focused on building the core consumer experience and the last one is related to all things monetization. Monetization is the revenue-generating org for Pinterest.
I lead the Machine Learning (ML) and Data Science teams within the Monetization org. Our customers are both Pinterest users and our advertising partners. In this day and age, a lot of the information matching is done using machine learning, and our job is to understand what users and advertisers are looking for and do the matching. We do our job well when we match the best ads to the interests and intent of our users.
Our work in Monetization ML is critical to supporting our users, advertisers, and our business. Relevant ads means a better experience for our users, higher ROI for advertisers and more money we can invest into the business to continue this flywheel.
I’m really excited about the advancements in the machine learning world and how they can be applied to our work at Pinterest. With large models predicting outcomes much better, we see a lot of opportunity in supporting users’ control over what they see, respecting users’ privacy choices and helping them through the journey from inspiration to realization (like connecting them with the most relevant ads). In this evolving, privacy-centric world where we need to thread together personalization in advertising and respect of user choices, this area of developing and applying machine learning models for advertising is really challenging (and fulfilling).
What led you to joining Pinterest?
When reflecting on previous roles and what led me to Pinterest, I feel midsize companies are in a unique spot to both be able to move fast but also have considerable impact in the world. My dream job is always building a small agile group of top technical talent who take on big product problems, move fast and create value for our users — a startup experience but large scale product impact. Pinterest is a place where people can really advance their careers by working with smart people in a collaborative way while learning a lot and taking their careers to the next level.
Pinterest has a very unique culture. Whenever there’s a problem, you get a lot of openness from various teams to work together and solve these problems. At bigger companies or organizations, energy is often spent on creating alignment across orgs to solve problems. At Pinterest, it happens naturally. When a challenge arises, cross-functional teams are very open and eager to help the team that raises the issue. It makes you feel very supported. This is also part of the secret recipe of Pinterest being able to move fast.
For me, it was also important to learn that Pinterest prioritizes a diverse and inclusive culture. I felt that our workforce is a role model for the rest of the industry even before I joined. During my interview process, I met with senior leaders across ML that emulated the type of culture Pinterest has, which was collaborative and inclusive by nature. Some of our most senior data scientists and ML engineers are incredible women who I admire and learn so much from each day. This is one of the reasons I was very impressed by Pinterest. I don’t think that these things happen by luck; it shows strong cultural values. I want to note, it’s not something I take credit for as I only recently joined, but it’s something I’m really proud of.
What makes Pinterest Engineering different?
For companies of our size (mid-size companies), we have some of the best ML infrastructure in the industry and some of the most advanced ML techniques. Companies that do the type of ML we do are usually much larger than Pinterest. Big companies are operating at this level, but they have hundreds of thousands of engineers. While at Pinterest, everyone here has a big scope and creates a high impact within our product and across our company. What really sets us apart is both the advanced techniques and technologies and being a midsize company where everyone has a big impact.
What would you say to someone who’s considering joining the Pinterest team?
People
The first reason why I join any team is the people. Our team has some of the smartest engineers and top industry experts in the field of ML, recommender systems, and product data science. Nevertheless, we have also managed to keep a collaborative culture, and everyone you encounter is very nice and welcoming. Oftentimes, when you operate at this level or height of tech, it can be competitive. This type of collaboration and genuine connection is rare to find, but you’ll immediately spot it when you join Pinterest. We recently interviewed a senior ML leader for a role at Pinterest. They emailed me afterwards that their interview at Pinterest was the most technically challenging interview they have done but also the most welcoming. It put a smile on my face. That’s who we are.
Product
What sets us apart from our peers is the positive impact of Pinterest on people. Every minute spent on Pinterest is in service of that moment of inspiration for our users. You can see the continuation of that commitment in our recent announcement to support the Inspired Internet Pledge.
Impact
You want to be somewhere you can have a lot of impact. There’s a lot of headspace and greenfield to do high impact work here. The size of the team is very small, so every individual and their work makes a significant difference to our product. As a result of our size, there is a lot of velocity in our org, and we move fast.
Tech and Science
We’re in the ML space. People want to work on the most innovative tech. We are one of the few mid-size companies with a good foundation and advanced ML technologies. We also have a very engineer-driven culture. Engineers have a lot of space to innovate and lead projects. Here, you can learn and apply the latest techniques across large models, Reinforced Learning, user representations and embeddings, user sequence modeling, privacy ML, and marketplace design. On the data science side, we are shaping the future of our product by taking on challenging user/product understanding work, causal inference through experimentation and other non experimental techniques and two sided marketplace analysis.
When you think about current engineering trends, which ones are you most excited about?
Large models in user understanding and recommender systems
There have been many advances in large language models resulting in a variety of techniques to train and serve these models. Those techniques and advances are now making their way into recommender systems and personalization of consumer products, so it’s exciting to see how this will translate to better personalization of ads, consumer products and evolution of recommender systems in the future.
Multi Task models and their extensions to large lattice models
In the old ML world, you would design a model for each individual task. Some of these technological advances allow us to combine models and have these bigger models address multiple tasks leading to more efficiency and generalization of user behavior.
ML and scaling processes across the company
Companies like Pinterest typically use a mix of human reviews and automated systems to (1) proactively identify policy-violating content, and (2) review/remove content that was flagged by users (for example, an ad that violates policies). With recent advances in ML, technology will be able to do more — scaling tasks and creating efficiencies — which ultimately helps free up human review to focus more on complex, strategic issues.
Generative AI
I’m looking forward to Generative AI trends — how GenAI can be used for engineering productivity and how it can enhance the user experience.
Anything else you’d like to share?
One thing I’d like to add is about the uniqueness of Pinterest’s product. Pinterest is a full funnel product. We take Pinterest users from the moment of inspiration through the moment of execution through shopping. Advertising and shopping is an intrinsic part of the core product. With other platforms, it doesn’t always feel as authentic. At Pinterest, the ad is in service of what the user is set out to do. That’s why at Pinterest, ads are in service of the user experience.
To learn more about engineering at Pinterest, check out the rest of our Engineering Blog and visit our Pinterest Labs site. To explore and apply to open roles, visit our Careers page.
