M.Sc. Data Science | AI & Machine Learning Research
My research interests lie at the intersection of Trustworthy AI, LLM Security, Causal Machine Learning, and Reinforcement Learning, with a particular interest in understanding and evaluating the behavior and reliability of foundation models.
I am currently completing my M.Sc. in Data Science at the University of Naples Federico II, Italy, and pursuing PhD opportunities in AI and Machine Learning.
- Trustworthy AI & AI Safety
- Large Language Model Security & Evaluation
- Causal Machine Learning
- Reinforcement Learning
Supervisor: Prof. Roberto Pietrantuono
University of Naples Federico II
My thesis investigates whether causal structure discovered from adversarial interaction trajectories can improve reinforcement-learning-based LLM security testing.
The experimental pipeline combines PPO-based adaptive testing, causal factor discovery, Fast Causal Inference (FCI), Partial Ancestral Graphs (PAGs), and structural causal modeling.
Selected results:
- Increased mean attack success rate from 19.33% to 61.54%
- Approximately 3.18× improvement over the RL baseline
- Reduced training-time API overhead by approximately 41.8%
- Evaluated zero-shot transfer to a different model family
Machine Learning: PyTorch · scikit-learn · Hugging Face Transformers
Causal ML: FCI · PAGs · Structural Causal Models
Reinforcement Learning: PPO · Stable-Baselines3
LLM Evaluation: adversarial testing · behavioral evaluation · safety evaluation
Engineering: Python · Docker · Kubernetes · Git
AI Systems Engineering Intern — RESTART / 5G Academy
Worked on a conversational AI system integrating automatic speech recognition, language-model-based dialogue components, and containerized services for edge/5G deployment.