Fabrizio Silvestri

Fabrizio Silvestri is a Full Professor of Computer Science at Sapienza University of Rome, where he leads the RSTLess research group on Robust, Safe, and Transparent Machine Learning and serves as Coordinator of the Ph.D. Programme in Data Science. His research spans Machine Learning, Natural Language Processing (with applications to Information Retrieval and RAG), and Generative and Agentic AI, with a recent focus on the reliability and interpretability of large language models.

Before joining Sapienza, he spent several years in industrial research. At Facebook (now Meta), London, he first worked on Facebook Search, where he productionized the query recommendation system serving billions of users daily; he then joined Facebook AI Research (FAIR), developing models to detect and limit the spread of misinformation across Meta's platforms. Earlier, at Yahoo Labs (London and Barcelona), he worked on web search quality, query understanding, and sponsored search, and before that he was a Senior Researcher at ISTI-CNR in Pisa.

He has authored nearly 200 peer-reviewed publications — more than 60 at CORE A/A* conferences (SIGIR, WWW, KDD, CIKM, WSDM, RecSys, ACL, EMNLP, NeurIPS, IJCAI, ICDE, ICDM, CVPR) and nearly 30 in Scimago Q1 computer-science journals — and his work has accumulated over 11,000 citations on Google Scholar. He regularly serves on the organizing committees of international conferences and workshops, including roles as General Chair, Program Chair, and Area Chair at leading venues in machine learning, natural language processing, and generative AI.

I coordinate the Ph.D. Programme in Data Science at Sapienza University of Rome. I currently teach the following courses:

• Applicazioni Informatiche del Machine Learning IT

  Applications of Machine Learning — taught in Italian.

• Generative AI EN

  Graduate course on generative models, LLMs, and agentic AI — taught in English.

• NLP Module — Natural Language Processing and Text Mining EN

  Module on modern NLP within the joint Natural Language Processing and Text Mining course.

I regularly supervise MSc and PhD theses on topics including retrieval-augmented generation, interpretability of large language models, graph neural networks for recommendation, and robustness of sequence models. Prospective students are welcome to get in touch.