AI Helps Students ‘Take the Hint’

Ishan Abraham BA ’26 is used to tackling big ideas. The San Francisco native is majoring in computer science and mathematics as well as minoring in data science and entrepreneurial leadership and innovation. Now his work has led to a major milestone: He’s the lead author for a paper that’s been accepted to the 21st International Conference on Cyber Warfare and Security, an international gathering of scholars and professionals to be held in Wilmington, North Carolina, in March 2026.

The paper, titled “SLMs Meet GraphRAG: A Structured Approach to Context Aware Cybersecurity Hint Generation,” presents improvements to a system that uses artificial intelligence to give students hints as they work through hands-on cybersecurity exercises. Instructors review the AI-generated hints prior to releasing them to students.

Jens Mache, professor of computer science
Credit: Copyright, Steve HambuchenWorking with Jens Mache, professor of computer science, and researchers from Northeastern University and The Evergreen State College, Abraham helped develop and test the system to address a common challenge in education: what happens when students get stuck. In many environments, wrong steps or missing pieces of information can derail progress and create frustration.

The new system, developed as a custom GraphRAG, is designed to provide helpful guidance. GraphRAG, short for Graph Retrieval-Augmented Generation, is a type of artificial intelligence system that combines two powerful ideas: a knowledge graph (a map of how ideas and information connect to each other) and a language model (an AI that generates written responses). Instead of pulling answers from plain text alone, GraphRAG searches through connected concepts and relationships, allowing the AI to give smarter, more accurate answers—especially for complex, multistep questions.

Ishan Abraham BA ’25 (left) and Professor Jens Mache“This system isn’t replacing teachers,” Abraham says. “Instead, it’s helping them respond to more students more quickly. Students no longer need to wait for help—the approved AI-generated hints point them in the right direction and help them think through the problem.”

The research team asked cybersecurity instructors to compare the custom GraphRag with an existing hint-generation tool. Instructors overwhelmingly preferred the new version, saying the hints were clearer, more accurate, and more useful for student learning.

Abraham’s role as first author on a paper accepted to an international cybersecurity conference is a rare distinction for an undergraduate researcher. At many conferences, papers are typically written by faculty, industry experts, or graduate-level researchers. Abraham’s conference attendance will place him in conversation with leading voices in the field. This underscores the power of hands-on, faculty-mentored research to prepare students for advanced study and promising careers.

Computer Science