👋 Hi, I’m Matteo Boffa, Assistant Professor at Politecnico di Torino and a member of the SmartData group.
I am a researcher working at the intersection of AI, cybersecurity, and language technologies, with a focus on building intelligent systems that can reason, adapt, and assist in complex security scenarios.
🔬 Research Focus
My recent work explores the use of autonomous and semi-autonomous agents for cybersecurity tasks, where large language models and agentic systems are leveraged to support threat analysis, detection, and response. I’m particularly interested in how these systems can move from passive tools to active decision-makers in security pipelines.
Alongside this, I have contributed to research on model interpretability, including work such as “The Sweet Danger of Sugar”, which highlights subtle risks in seemingly benign model behaviors and emphasizes the importance of understanding how models reason and fail.
🧠 Background
I come from a background in log analysis and anomaly detection, where I applied NLP techniques and language models to extract structure and meaning from large-scale system logs. This experience shaped my current approach: treating security data as language, and security problems as reasoning tasks.
🚀 Interests
Agentic AI systems for real-world applications
LLMs for cybersecurity and threat intelligence
Interpretability and trustworthy AI
Bridging NLP and security analytics
📫 Get in touch
Don't hesitate to contact me on my personal email or linkedin profile!