MSAI Student Uses Reinforcement Learning to Improve Customer Service Strategies at Capital One

During the summer months, students of the MS in Artificial Intelligence (MSAI) program spend the term off-campus completing their degree-required industry internship. During this experience, it is important for students to exhibit their knowledge as well as demonstrate practical application using a variety of skills in a real-world, professional setting.

We caught up with current MSAI student, Nico Tyjeski, to hear about his internship at Capital One. Nico was proud to talk about his experience, the work he is doing, and career plans once he finishes his graduate program at Northwestern in December 2019.


Q:  What department at Capital One are you working with over the summer? 
A:  Card ML

Q:  What is your title?
A:  Data Science Intern

Q:  Is the work a valuable experience in relation to your academic studies? If so, why?
A:  Absolutely. This internship has pushed me to become a better programmer and learn a variety of tools that I think will come in handy in the fall. Working with seasoned data scientists and software engineers has taught me a lot about building end-to-end applications and managing projects, and spending the summer working on a reinforcement learning project has meant reading plenty of research papers and developing even deeper understanding of the topics covered in our classes.

Q:  Describe the responsibilities that enabled you to apply the knowledge and skills you are learning through your MSAI coursework to your role.
A:  My work this summer has involved researching, developing, and deploying a reinforcement learning agent that continuously optimize the strategies used in call centers. Along the way, I have been responsible for investigating the benefits and best practices of hosting RL agents on EC2 Kubernetes clusters.

Q:  Were you allowed to take the initiative to work beyond the basic requirements of the job? If so, in what way?
A:  Yes, my supervisors have allowed me to expand upon the project when I see opportunities to add value. By creating a custom environment to simulate specific customers and processes, I was able to test a variety of agents and agent models under a variety of challenging conditions. This has given us insight into model risks and made clear the pros and cons of different RL solutions. I also expanded the way we approach the problem to capture relationships between customer features and varying system dynamics.

Q:  How and in what capacity did the organization and/or your supervisor work with you regularly or support your learning?
A:  My internship has been framed as being part “Modeling” and part “Platform”. I have two supervisors, each specializing in one component of my project, who act as my sources of regular guidance and feedback. The Data Science teams also put on regular Lunch & Learns; the most recent one I attended ended up being especially relevant to my work: “A Deep Dive into Bayesian Modeling”. 

Q:  Briefly note new skills, techniques and/or knowledge you gained during your internship.
A:  Over the course of this internship, I have had the opportunity to work with and learn a great deal about Multi-armed contextual bandits, Markov decision processes, Monte Carlo methods, Bayesian modeling, Flask, Docker, and Kubernetes.

Q:  What has been your favorite experience or the coolest thing you were able to accomplish?
A:  All of Capital One’s Data Science interns from around the country were flown to Austin, TX for this year’s SciPy Conference. I had the opportunity to meet many brilliant people, learn all about the open source Python community, and eat some of the best smoked brisket in the country.

Q:  In what way do you believe MSAI prepared you for this internship?
A:  The MSAI curriculum has been very good about covering both the modeling and the platform side of DS/ML/AI, much like the way my internship has been framed. While I have learned a great deal this summer, I’ve been fortunate that almost all topics had been previously introduced to me in some capacity through a MSAI lecture or talk.

Q:  You will graduate from the MSAI program in December. What’s next for you?
A:  My hope is to become a full-time Data Scientist or ML Engineer!


The MSAI program offers students access to corporate connections and internship opportunities through its esteemed Partner Program. Likewise, as with Nico, students are also able to identify prospects matching their interests through successful networking and development of their own personal and professional connections.

The Master of Science in Artificial Intelligence combines four quarters of rigorous coursework with one additional term dedicated to the internship experience. The curriculum—designed for professional development—delivers hands-on training, equipping students with the knowledge and skills necessary to develop AI solutions using technologies of today. Learn more.