HDSI Industry-Mentor Capstone Fosters Benefits for Students and Employers

Story by:

• Jack Imel
• Kimberly Mann Bruch- [email protected]

Share This:

Article Content

Organizations constantly have data-related problems that need to be solved to help with business objectives, as well as solving societal issues involving topics such as energy, data ethics and policies. Since it was founded in 2018, Halıcıoğlu Data Science Institute-a part of University of California San Diego's School of Computing, Information and Data Sciences (SCIDS) - has been providing an opportunity for students and industry leaders to come together and tackle the growing demand for data science and AI innovation and utilization.

For years, HDSI has collaborated with industry partners like Prism Data and San Diego Gas and Electric (SDG&E) to spur innovation, train students and even develop talent pipelines through its Senior Capstone Program.

"Data science is still a rapidly evolving field where the book is still very much being written and this is quickly being accelerated by the emergence of AI," said HDSI Visiting Professor Umesh Bellur, who is the Senior Capstone Program's faculty chair. "It is more critical than ever that there is a close connection between industry and academia as we continue laying the foundation for the discipline while at the same time drive new practical applications."

Building Better Credit Assessment Tools with Prism Data

Brian Duke, former vice president of data science at Prism Data - a next-generation credit scoring and financial data analytics company founded in 2021 - explained the importance of the program for students and industry partners alike.

"Undergraduate courses typically do a fantastic job of teaching students 'what' to do and 'how' to do it," Duke said. "There is, however, a third aspect that is vital to learning, but that often receives less attention -'why' should we do it?"

Prism Data has been engaged with the HDSI Senior Capstone program for three years. During that time their team has met with students on campus, helped them practice the techniques they're learning at HDSI in real-world scenarios and effectively created a talent pipeline for their organization.

Kyle Nero, an HDSI alumni who participated in Prism Data's first cohort, was offered a full-time opportunity and has now helped mentor two cohorts through their capstone experience. For Prism Data, working with talented students like Nero is a big part of the "why" partnerships between industry and education play an important role in achieving business and societal goals such as better credit evaluation.

"Staying at the forefront of innovation is critical for companies like Prism - and that means continually being involved with cutting edge research, approaches and talent," Duke said. "These partnerships enable Prism to work with top-notch students and experiment with approaches while maintaining educator-level relationships to potentially collaborate more fully in the future. The opportunity to engage with multiple students in the later years of their undergraduate studies builds a growing pipeline of talented individuals who either end up working at Prism, or being aware of Prism - magnifying the company's reach."

Duke and Nero recently worked with UC San Diego undergraduate students using Natural Language Processing (NLP), a subfield of artificial intelligence (AI), to broaden the scope of credit assessment by including individuals with limited credit history. To do this, the students trained NLP models on multiple datasets that provided de-identified consumer transaction details, account balances and delinquency indicators - creating a "cash flow-based credit score" that was shown to predict delinquency better than a traditional credit score. With an end goal of creating a more holistic credit assessment while maintaining accuracy and transparency, Duke made sure the vital "why" was at the forefront of the students' attention while working on the project.

"We have made it a goal of ours to allow the question of 'why' to remain prevalent throughout each lesson of the six-month project," Duke said. "When students have the opportunity to occasionally step away from the raw data and programmatic modeling and really think about the purpose behind a project, it enables them to uncover insights that may have been very hard or even impossible to find mathematically, and improve their critical thinking and problem solving abilities which can eventually be applied to other real-world problems."