Alessandra Terranova

Research keywords: Natural Language Processing, LLM-based Agents, Experiential Learning, Retrieval Augmented Generation, Misinformation Countering

Bio:

Alessandra is originally from Cosenza, Italy, and she moved to Edinburgh in 2020 to start university. She has an interdisciplinary background bridging Linguistics, Cognitive Science, and NLP, with a BSc in Cognitive Science from The University of Edinburgh.

Her first encounter with NLP research was through her undergraduate dissertation on Retrieval Augmented Generation for the summarisation of dialogue, and she has not stopped exploring the field since.

Alessandra has applied her expertise in both industry and research settings: at Aveni, an Edinburgh-based fintech startup, she worked as a Junior NLP Engineer, developing AI solutions tailored for the financial services sector; she also contributed as a research assistant on the UTTER project at The University of Edinburgh, focusing on memory for conversational AI technologies.

Most recently, Alessandra interned as a Site Reliability Engineer at Google Dublin, where she employed LLMs for the classification and analysis of incoming bugs for the Cloud BackupDR team.

PhD research:

Alessandra’s research focuses on developing safer and more responsible large language model (LLM)-based agents by integrating human-like metacognitive mechanisms such as intellectual humility and perspective seeking. She is particularly interested in the design of NLP systems that can learn from experience and collaborate effectively with humans, ensuring robustness and fairness in complex, real-world applications. Her work explores the use of retrieval-augmented generation and memory mechanisms to improve agent trustworthiness and adaptability. Alessandra aims to study the application of these systems to social challenges, including content moderation and countering misinformation, with an emphasis on human-centered design.

Supervisors: Alexandra Birch-Mayne, Björn Ross