USF students and faculty from the Bellini College of Artificial Intelligence, Cybersecurity and Computing participated in the first CyberBay 2025 conference, an event that brought together technology, academic, government, and military sectors to promote Tampa Bay as a center for cybersecurity, artificial intelligence, and national security.
Graduate student Dayne Guy presented research on training large language models to recognize obscure or figurative language. He said, “I’m an AI student but it’s been very cybersecurity focused, which I appreciate because I don’t really get to see much cybersecurity in my day-to-day classes that are more computing and AI. I also like hearing from the different speeches, the keynotes. There are some great takeaways from this, and you’re also learning about a lot of interesting research.”
USF’s CyberHerd teams competed in a Capture the Flag (CTF) challenge at the conference. Jun Lu described his experience working on a task involving network spoofing and setting up a man-in-the-middle attack proxy: “Right now, I have two jobs I’m juggling. Basically, in this network, if you connect to it, there’s a client. I’m not sure what the exact IP is. It’s going to transmit a message to this. So, the idea is that you’re supposed to spoof the IP of it, and also set up something called a MITM (man-in-the-middle attack) proxy. It’s a tool. The problem is that I am trying to off memory on how to set this up because I haven’t done this myself before.”
Although USF fielded multiple teams in the competition, Squid Proxy Lovers—a team with members from several universities—won the $20,000 grand prize. Runners-up included UCF Knighsec B Team and U.S. Air Force Academy.
Nathan Leung of USF CyberHerd explained their approach: “That’s the best bet. Then you have a team of four different skillsets. You’re just figuring out who does what the best and it varies by person. Like, for me, I’m into open-source intelligence and forensics. I definitely think I do that the best.”
Students also participated in cyber debates on topics such as teaching ethical hacking methods and whether universities should prohibit generative AI tools like ChatGPT in academic writing. The debates were organized by Professor Sriram Chellappan.
Michael Maldonado Cruz reflected on his experience: “Looking at the whole experience, I think it was great,” he said. “It forces you to get out of your comfort zone. It’s a lot of fun coming out here, engaging in your communication skills and your ability to drive a point to convince an audience on a point you might not necessarily agree with.” Cruz argued against prohibiting generative AI tools in academic writing and was named Most Valuable Debater by audience vote.
Chellappan collaborated with an outside company for debate preparation: students received topics ahead of time but did not know what questions would come from judges or audience members.
Miguel Mateo Osorio Vela shared his thoughts: “I really think we need this type of events as computing professionals,” he said. “It got me to learn so many things about cybersecurity… It challenged me to convince my audience of what I think is correct. But I need to sustain my arguments with evidence… That was the most important thing that I learned through the coaching sessions.” He added that it was his first formal debate experience.
Throughout three days at CyberBay 2025, USF students and faculty presented research projects covering areas such as redesigning generative AI for children’s STEM-based writing support; event segmentation models; 3D imaging analysis; research challenges in cybersecurity; demographic alignment in large language models; improving security for vehicle networks; and grammar-based parsing using LLMs.
Long Dang presented work on an AI solution for autonomous vehicles: “I was able to talk to so many people from my industry,” he said. “I like to go around and challenge them on their claims… Also getting around to talk to other professors… is great… I want to become a researcher like that.”
The inaugural event concluded Wednesday with plans announced for another conference in spring 2026.
