Over the past few months, I’ve been diving into both Natural Language Processing (NLP) and Machine Learning in Cybersecurity, mixing theory with real-world applications. My work with Transformer architecture and fine-tuning in NLP, through practical labs and assignments, has been a key part of preparing for my master’s thesis on Log Anomaly Detection using open-source LLMs.

My passion for the math behind ML started during a Foundations of ML course at Camerino, where I focused on the theory of machine learning, digging into the algorithms and mathematics behind ML models. This gave me a solid theoretical base.

Then, here at Reykjavik University, I honed my practical skills through the Machine Learning in Cybersecurity course. We tackled real-world issues like malware detection, network intrusion, and spam detection with machine learning models. One of the coolest moments was a competition to build the best-performing malware analysis model—I was thrilled to win with the best F1 score. That challenge really showed me how to combine theory with hands-on application.

We also dove into AI’s limitations, especially around adversarial attacks—a reminder that AI systems in cybersecurity need to be resilient and robust.

These experiences have strengthened my understanding of AI, from the math behind it to applying it in cybersecurity. Moving forward, I’m excited to focus on bridging the gap between AI and real-world security challenges, especially with my thesis on Log Anomaly Detection.

Check out the stuff I’ve been working on: