Meet the Tinder Machine Learning Team | Steven Brotz
There are many teams at Tinder that work together to create meaningful experiences for our members. In this series, we’re taking a behind-the-scenes look at the Machine Learning team, in order to find out more about the people behind the product. Meet the Machine Learning team.
Machine Learning Software Engineer Steven Brotz first arrived at Tinder as an engineering intern in the summer of 2018, before his senior year of college. After graduation, the unique environment that the Tinder app provides for curious engineers drew him back in full time. Since then, he’s focused on many projects that contribute to the safety of Tinder members by detecting ‘bad actors’ in the app.
We chatted with Steven to learn more about what working in Machine Learning at Tinder is like.
First, could you tell us three fun facts about yourself?
- I like playing Super Smash Bros. Melee on GameCube
- I play guitar and I love music
- I hate chess, but I still play it all the time
What drew you to work at Tinder?
I actually started at Tinder as an intern in 2018 where I was really fascinated by the product, as well as the clearly challenging work required to make it work the way that it does. I greatly enjoyed the culture of my team and the company, and so I made the decision to join full time.
What’s a ‘day in the life’ as an engineer at Tinder?
It depends on where I am in the life cycle of my current project. In the early stages, I may be testing the viability of some ideas in a notebook. Some days or weeks later, I’ll be engaging in deep work, trying to extract a training dataset from big, sometimes noisy data. Then, I’ll be training and tuning the model, and then working hands-on with backend and infrastructure engineers to deploy it at scale in production. Throughout the project, I’m always communicating with my teammates every day, and with project stakeholders regularly.
What are some accomplishments that you’re proud of in your career?
I’m proud of creating a Natural Language Processing (NLP) model that’s responsible for taking down bad actors in the app before they can do harm to genuine Tinder members. I’m also proud of the ‘Rule Health’ dashboard, which lets our team monitor the performance of our anti-bad-actor efforts in terms of accuracy, volume, user outcomes, and other metrics.
What excites you most about your role?
I enjoy the cat-and-mouse game that is anti-bad-actor work. We’re always trying to stay as many steps ahead as possible, and trying to make it difficult for bad actors to realize both the fact that they’ve been detected and what we used to detect them.
How has your experience at Tinder shaped your career as an engineer?
Tinder is my first non-intern position, and I’ve had consistent opportunities to develop my skills in Machine Learning modeling; big data Extract, Transform, and Load (ETL); project ownership; deployment; monitoring; and engineering in general. It’s been great to start my career in a place where I can get real experience doing so many different things.
What lessons or growth opportunities have you experienced in your role?
I’ve learned the benefits of focusing on practical, impactful solutions, as well as seeking simplicity over complexity as much as possible. Since the time I was a new grad about two-and-a-half years ago, I’ve been given so much room to grow: to make more decisions, to create plans, to weigh trade-offs, to lead discussions, and more.
What might surprise engineers about working at Tinder?
The number of Trust & Safety ideas that come directly from engineers using the product and noticing things we could do or make better might be surprising. Many bot patterns have been first noticed by an engineer swiping on the app.
Interested in solving problems and working with a dynamic team of engineers? We’re hiring!