Deciphering Document Data at Capital One

Nico Tyjeski (MSAI '19) recognizes his field might not sound exciting at first.

"To an outsider, I’m sure processing documents sounds pretty dry,” he said. 

To him, though, his role as a data scientist at Capital One is anything but that. 

“My team uses cutting-edge deep learning models to solve human problems in creative ways," he said. “We have an incredibly positive and collaborative culture, and I’m constantly given the opportunity to explore and learn new things."

Tyjeski first started working at Capital One as an intern in 2019 while still a student in Northwestern Engineering's Master of Science in Artificial Intelligence (MSAI) program. He joined the company full time in January 2020 and today is a principal data scientist on a team focused on document intelligence. His main focus is streamlining and automating operations around the company where incoming documents would otherwise need to be reviewed or processed manually. 

To do that, he uses artificial intelligence that is able to read documents, understand them, and either take action or make recommendations.

“My team manages a platform where computer vision and language services are used to process documents and provide users with whatever it is they need to know about their content,” he said. 

To grow that platform, Tyjeski relies on the knowledge he gained as a student in the MSAI program. 

“Every time I build a model or run an experiment, I leverage a range of technical skills learned in the MSAI program,” he said. “When collaborating with teammates, I use experience gained through the program’s capstones. When participating in technical reading groups, I apply lessons learned conducting research and attending similar groups in MSAI.” 

These types of practical, industry-relevant lessons were exactly what Tyjeski wanted out of a postgraduate education. After earning his undergraduate degree from Northwestern University in industrial engineering, Tyjeski chose MSAI in part because of the exposure he had to the program. 

“I was already familiar with the professors and the quality of instruction,” he said. “I was drawn by the specialized focus, industry tie-ins and the opportunity to study alongside those with similar career goals.” 

While the study of artificial intelligence requires some very specific technical knowledge, Tyjeski found one of the program’s most valuable lessons did not come from a book or lecture.

“Ironically, the most important lesson I learned was how to teach myself,” he said. “State-of-the-art in deep learning moves at a fast pace, so any particular model a professor teaches in class may not be relevant by the time you start working. To address this, we were taught how to read the latest research papers, judge them critically, and implement the approaches ourselves.” 

These skills worked well for Tyjeski, and led him to discover the unexpectedly colorful career he has today.

"To a data scientist, the work is pretty similar," Tyjeski said, "whether your model is for tax forms or tweets, drivers licenses, or pictures of puppies."