Students collaborate with Bloomberg Industry Group on 'messy' legal industry challenge
Graduate students used machine learning to improve classification processes for complex legal data.
From left: Marianne Bellotti and Jacques Marais from Bloomberg Industry Group and Raphael Esedebe, Jay Sarode, Alip Yalikun, and Jacob Albright from Virginia Tech. Photo by Lydia Fahey for Virginia Tech.
When Bloomberg Industry Group needed new ways to navigate the complexities of legal data and find connections between sprawling regulations, court decisions, and contracts, its team turned to Virginia Tech.
The group partnered with a team of Master of Engineering students through the program’s project-based learning capstone, which provides students the ability to address real-world problems before they graduate. Together, the students and company representatives tackled one of the legal industry’s most pressing challenges: how to extract, classify, and connect citation data buried in long-form, unstructured legal text.
The name of the project? LawGorithm.
“We’re not just building something that goes ‘beep boop,’” said Raphael Esedebe, one of the students who worked on the project. “We’re working with content that has meaning, that someone could actually read and use to make a decision. That’s powerful.”
Capstone projects are a component of the Master of Engineering program’s project-based learning model. The university intends for project-based learning to make up between a third and a half of the curriculum. Student teams have previously worked on projects with firms such as The Boeing Company, RTX, and Torc Robotics, and organizations such as the U.S. Marine Corps. These courses are designed for students to build real-world skills within the comfort of a collegiate environment so when they graduate and begin their first job, they are confident and prepared.
For Esedebe and fellow students Alip Yalikun, Jay Sarode, Evan Lee, and Jacob Albright, the project was a hands-on introduction to legal tech and natural language processing as well as an opportunity to engage with professionals in the field.
Partnership and mentorship
Students said they found value in working with Bloomberg Industry Group mentors, especially Marianne Bellotti, director of engineering. The students met weekly with Bellotti, who offered guidance and expertise on the project as needed. The students were also connected to other artificial intelligence and machine learning experts at Bloomberg Industry Group, like Jacques Marais, a senior software engineer.
Esedebe appreciated the balance of professional expectations with human connection and understanding. “They don’t hand-hold you, but they are very accommodating and understanding. Maybe that’s not the experience with every industry, but with Bloomberg Industry Group, that’s been a really big benefit.”
Yalikun echoed the sentiment: “I learned so many things that I couldn't have learned in a classroom. Our instructors are great, but this kind of experience adds something else. I loved that we worked on a real project that is valued by a company.”
The technical challenge
Bloomberg Industry Group wanted to explore the interaction of laws, regulations, and court cases from a graph perspective. The students got to work, training and testing a natural language processing (NLP) model that could make sense of long, complex legal documents packed with critical information – such as what laws were mentioned (referred to in the research as “referenced laws”), when the laws took effect (“effective dates”), and how documents were categorized (“document types”).
Albright was tasked with the data visualizations. In the beginning, he said, they didn’t have trained models or real data. Bellotti introduced him to stubbed data, or placeholder data, that is not fully accurate but good enough to allow the team to move forward with the project.
“Using this placeholder data made such a difference, especially in a project with so many moving parts,” Albright said. “It let me show something tangible to teammates and mentors. Even if it wasn’t final, it was something we could work from.”
The students worked like a professional software team by testing ideas, integrating feedback and more complete data as the NLP model began to function, and refining weekly.
Sarode said it was exciting to be part of a project with so much potential to grow and evolve. “If a new student were to pick it up next semester, they could take it in a totally different direction,” he said. “But we were laying the foundation.”
Iterative and team-based learning
The students said the capstone offered a more realistic simulation of how software development works in industry compared to typical academic assignments.
“In academia, a lot of times you’re just cramming, and then shoving it all together at the very end and hope for the best,” Esedebe said. “Whereas now [with the capstone], it’s a very iterative process. It feels like we’re going through stuff for a reason.”
With consistent guidance from Bellotti, the team was able to reframe its approach to a more iterative one by integrating early and often.
“We’d been focused on perfecting the model,” said Sarode. “The Bloomberg Industry Group team helped us see that progress comes from getting something working, even if it’s imperfect, and improving from there.”
That advice paid off. As the students rotated leadership roles and split the workload, they discovered the importance of staying connected and informed across the entire project.
“It’s not just enough to be an expert in what you’re working on, you have to be an expert in what the entire group is doing,” said Esedebe. “You don’t have to build it from scratch, but having a fundamental understanding and being able to communicate it is so integral.”
The benefit of industry partnerships
For Virginia Tech, partnerships like this one offer industry-aligned educational value and a pathway to workforce readiness for students.
“Experiential learning is the heartbeat of our master’s program,” said Melissa Cameron, assistant collegiate professor of computer science and core faculty member of the newly formed Virginia Tech Institute for Advanced Computing, based in the D.C. area. “These projects let students step into real-world roles, apply their knowledge in meaningful ways, and build professional relationships that often extend beyond graduation.”
For Bloomberg Industry Group, the collaboration offered fresh thinking on a complex challenge, all while building goodwill and connections with future tech leaders.
“We were excited to see how the students would approach a real, messy, and very human problem,” Bellotti said. “What really stood out was that the students brought a people-first mindset to the project. They weren’t just focused on building the most technically sophisticated model – they were thinking about how to make it useful for the people who rely on our platform every day.”