Senior Project Builds Real-World AI Tool
The group that created Lobster AI presented at the senior showcase at CCI.
Over the course of an academic year, a group of Drexel University students from the College of Computing & Informatics (CCI) developed a new artificial intelligence (AI) tool that can help streamline the medical writing world. To accomplish this, they worked with industry partner AOIC, LLC, a medical and scientific communications company, and got further than most expected they would in a senior project.
“It was an ambitious project that the students made a tremendous amount of effort on,” said Mark Altieri, stakeholder company AOIC’s vice president of finance and technology who gave one group of students direction on the technology requirements. “The thing that impressed me the most after their technical expertise was that we had a really dynamic group that each brought expertise to the table. I never saw an unprepared student on a call. They all were ready to present the things they'd worked on and had obviously put in the hours in between the meetings."
Seniors in CCI develop software projects through a year-long class, working usually with real-world, professional partners to develop software projects. This means everything from taking the process from budgeting and planning all the way through to design and implementation. Sometimes, the projects are too in-depth to be fully completed in the nine months they have, but it’s as close to a real-world project as you can get, adding to the experience students get on co-op. For students who are at Drexel for five years, they get three six-month co-ops where they work in their chosen fields while still learning.
“If you add the 18 months of co-op to nine months of senior project, you're walking out of school with 27 months of experiential learning, and I don't know any other universities that are doing that,” said Jeff Salvage, teaching professor of computer science.
The group of students included Leo Li, software engineering ’25; Abby Tabas, computer science ’25; Maame Adwoa Ocran, computer science ’25; Jason Morais, software engineering ’25; and Viven Ho, software engineering ’25. They were tasked with using AI to help develop a first draft of a medical manuscript, which is a formal document that details the results of a medical trial or research. They include a method section, which explains how a study was carried out, including research design, data collection and data analysis. This can be time-consuming for writers despite its structured format. Sarah Hummasti, AOIC’s vice president of scientific strategy and services who heads the medical writing team, said her team has wanted to find ways to streamline that process. For this project, they focused on phase three clinical trials, which test the safety and efficacy of a new treatment compared to previous treatments. There’s a protocol in the trial that lists everything about how the study was done, but that can be hundreds of pages.
“For a manuscript, when you want to share it with the rest of the scientific and medical community, you need to condense that to something that provides all the key and important information,” Hummasti said.
The project was complicated by security and confidentiality protocols around the data that AOIC stewards, so the students worked with data that was already in the public domain. It was real-world data, Altieri said, but from a manuscript that had already been written.
Hummasti provided examples of method sections and a checklist that can be used to create one, and the students trained the AI based on what the writer would do. Building the AI was a labor-intensive process because of the confidentiality concerns, and because while the manuscripts themselves are usually very formulaic, each trial is different.
“It gave them the starting point and the end point so they could use that to train their AI,” Hummasti said. “There are always going to be some differences, but what are the things that are consistent, and how do we home in on those so we make sure we’re capturing the key things?”
The students made progress weekly, updating AOIC representatives in meetings to talk about their approach and ask questions. Salvage prioritizes getting students facetime with the stakeholders of their projects, because it helps them develop the soft skills that are vital in client relationships, like communication, teamwork, professionalism and time management. Plus, they see what a real-world client experience is like, and that real-time, real-world connection helps students see what’s going on in their industry in the moment.
“I think that the atmosphere and the expectations that Jeff put in place are very important,” AOIC President and CEO (and two-time Drexel alumnus) Tom Burke said. “The class is clearly a serious, real-world focused class. The students’ participation and dedication are important, and you don’t achieve that unless the backbone of the program is built through the faculty and the college.”
Ultimately, the students developed a tool known as Lobster AI, and over the course of the year, the group got much further along than AOIC expected. As far as the AOIC team knows, this is a new product that hasn’t been developed to market by any other companies, so it gave the students a lot of experience on what kinds of innovation employers will be looking for as they head into the workplace.
“It’s a novel approach, and it might end up bearing fruit,” said Altieri, the vice president of finance and .
Now, the group of Li, Tabas, Ho, Morais and Ocran have graduated and are out in the real world. As for AOIC, they’ve matched with another group of seniors who can take the tool to the next level in the 2025–26 academic year. They know they can count on Drexel students.