DUCK Lab

Luca Andolfi is a Research Fellow in AI at the Imperial Policy Forum, within the Strategic Engagement Division at Imperial College London, where he delivers research and expert-led training on AI directly relevant to government policy. He is completing his Italian National PhD in Artificial Intelligence at Sapienza University of Rome, supervised by Prof. Marco Console, with a focus on knowledge representation and the informativeness of query answers over ontological knowledge bases. In 2024 he was a Visiting Fellow at Imperial College London, working with Prof. Eleonora Giunchiglia on neuro-symbolic AI. His research sits at the intersection of formal logic, knowledge representation, and machine learning, with particular interest in methods for learning semantically meaningful concepts that support interpretable and robust reasoning within neuro-symbolic architectures.

Mihaela Cătălina Stoian is a postdoctoral researcher at Imperial College London working on constraining LLMs for theorem proving. She recently completed her PhD at the University of Oxford, where her work focused on developing neuro-symbolic methods that integrate background knowledge constraints into neural networks for real-world applications. Her contributions to the field have been recognised with several awards, including the Oxford PhD Runner-up Prize awarded by G-Research. Previously, she worked on detecting reflective symmetries in 3D models at Five AI and completed her Master’s in speech-to-text machine translation at the University of Edinburgh.

Lohith is a first-year PhD student in the DUCK Lab at Imperial College London, supervised by Dr. Eleonora Giunchiglia. He holds an MEng in Information & Computer Engineering from the University of Cambridge and, prior to his PhD studies, worked on low-latency speech-language models for speech synthesis at a venture-backed startup. His current research focuses on the application of Neurosymbolic AI in Financial Markets, including constrained reinforcement learning for dynamic portfolio construction and efficient constraint enforcement over categorical variables with applications in foundation models for automated trading.

Joshua is a first-year PhD student at Imperial College London and a visiting researcher at the University of Edinburgh and Mohamed bin Zayed University of Artificial Intelligence (MBZUAI). Prior to his PhD, he published several papers at top-tier conferences, including ACL, EMNLP, NAACL, and ICLR. His research interests lie within LLM Reasoning, with a particular emphasis on theorem proving to improve the faithfulness and verifiability of LLM outputs. He is currently working at the intersection of diffusion models and theorem proving.

Dylan is a first-year Chemistry PhD student funded by the AIChemy Hub for AI in Chemistry and co-supervised by Prof. Kim Jelfs, Dr. Alex Ganose, and Dr. Eleonora Giunchiglia. His research focuses on neurosymbolic AI as a means of embedding prior chemistry knowledge into generative models for materials discovery, with the aim of improving the synthetic accessibility of generated structures. He holds a first-class MChem from the University of Oxford, where his Masters research, supervised by Prof. Fernanda Duarte, centred on the computational design of serine beta-lactamase inhibitors to combat antimicrobial resistance.

Yani Zhang is a final-year PhD student at the Chair for Mathematical Information Science at ETH Zurich, advised by Prof. Helmut Bölcskei. Her research focuses on neural network theory and formal many-valued logic. She received her master’s degree in Electrical Engineering in 2022 at the Technical University of Darmstadt, where she wrote her thesis with Prof. Marius Pesavento.