What does a data scientist do at Capital One?

At Capital One, data scientists are a vital part of creating products, improving processes and changing banking for good. They collaborate with partners across the company to build models and implement data in innovative ways. Their work spans the entire business; from bank to card to commercial to auto finance.

We spoke to a few of them about their work, day-to-day tasks and what candidates for a data scientist position can expect about life at Capital One.

What do you do as a data scientist at Capital One?

Hossein, senior director of data science and machine learning in Financial Services: At Capital One, data scientists are responsible for the predictive side of the wide range of analytics that the company does. As a senior director, I help my team solve problems and brainstorm new solutions for the models we’re using. I also focus on development, and make sure my team is able to learn and grow. I came from the world of academia, so learning and development is important to me.

Ali, director of data science in Card Data: My job is a mix of things. I do a lot of hands-on work with data, but I’m also making presentations for product partners as we build new customer experiences. I’m able to work in the very technical side of data and then see how that’s used across the company.

Kyaw Hpone Myint, principal associate of data science in Machine Learning: I have a Ph.D background, so I like trying new ideas. I spend time working on my main project, which is refining current Capital One machine learning models. I’m also able to take a few days to develop side projects and present them to leaders.

What projects have you worked on?

Kyaw Hpone Myint: I work on fine-tuning large language models. We have a lot of customers, and my work helps us track the customers throughout Capital One products and provide them with a better customer experience.

Ali: Right now, I'm working on a project called Block Charges, which is a subscription blocking product for customers who use capital One credit cards. We have two models the team has built. One is a model that predicts if a payment is part of a recurring series of transactions, and then we present that to the customer. We'll give them an option to block or cancel it. We’ve built a pretty great model for it, and we’re excited to put it into production.

Hossein: I’ve worked on many projects over the years. The very last project that my team has been working on recently is rebuilding our business’ risk models. When someone applies for a loan with us, these models are the very first step for us in deciding if we are able to approve or decline the loan. Assessing a consumer’s risk is not as straightforward as someone might think. It involves looking at a lot of past data, building various models and then assembling them together. Working on that project was very rewarding, because if I can make that model even a little bit better, it means that we would be extending credit to people who might not be served otherwise. 

How does Capital One encourage data scientists to innovate?

Kyaw Hpone Myint: Learning and innovation are the same. We have educational assistance, free learning opportunities and internal and external conferences. We have an internal conference for data science that lasts a week in Washington, D.C. You’re able to listen to people’s research and see the work that’s happening across the company, all while the cherry blossoms are blooming.

Ali: I've been involved with Capital One’s patent program, OnePatents, since 2018. This program rewards employees for submitting patent ideas. It’s a niche place within the company where a lot of innovation happens. Every year we do a brainstorming session, where we get together as a team and then split up into subgroups and submit patent ideas. I’ve filed over 30, and seen a handful of them go through the complete process. Capital One helps you make that investment and guides you through the process. It's really neat to see your name on a piece of history.

Hossein: Capital One was founded on the idea that if you leverage data and analytics, you can extend credit to more people. That central idea is still with the company to this day, and we really encourage innovation. We’ve made huge strides in the tools and data we use. We’re always asking how we can use data and data scientists to solve different problems across the company.

What else should data scientists know about Capital One?

Hossein: I’m passionate about solving hard problems, and I really enjoy helping others learn and grow. Through my school years, I always thought I needed to pursue a career in academia for that. As I worked here and got more comfortable in my role, I realized that I can do all those things that I wanted at Capital One. I can work on very hard problems and try to solve them. These solutions end up impacting and changing people's lives. And I realized I can help others develop and grow by sharing my experience and learnings with them.

Kyaw Hpone Myint: At Capital One, data scientist roles are more like applied scientist roles, where we’re able to research and experiment. We are not only building machine learning models, we apply them and are able to see the whole process. I can follow a model from beginning to end and see how it is applied in the business.

Ali: People think I work on very traditional finance work, but I’m more focused on digital products and experiences for customers. There’s more to data science at Capital One than the traditional view of what data science and finance might be.