Team

Sonali Parbhoo

Sonali Parbhoo is an Assistant Professor in the Department of Electrical and Electronic Engineering at Imperial College London, where she leads the AI for Actionable Impact (AI4AI) Lab. Previously, she was a Postdoctoral Fellow at the John A. Paulson School of Engineering and Applied Sciences at Harvard University, supported by a Swiss National Science Foundation fellowship. She received her PhD (summa cum laude) from the University of Basel, and her MSc from the University of the Witwatersrand, South Africa. Sonali has published more than 60 papers in machine learning venues such as ICML, NeurIPS, ICLR, AISTATS, KDD and JMLR, and in medical and scientific journals including Nature Medicine, The Lancet Digital Health, JAMA and Intensive Care Medicine. She has co-organised Machine Learning for Healthcare (2025, 2026) and Advances in Approximate Bayesian Inference (2024, 2025), as well as workshops at ICLR, NeurIPS and ICML, and serves as an Area Chair for ICML, NeurIPS, ICLR, RLC and MLHC. Her honours include a Rising Star in AI Award (EECS, Harvard, MIT 2021) and a CIFAR ELLIS Award. Her vision is to create actionable AI and ML systems that work with humans to improve scientific inference and predictive capabilities in healthcare, focused on challenges of reliability and robustness.

Dominic Marshall

Dominic is a Clinical Research Fellow and PhD candidate at Imperial College London, and a Specialist Registrar in Respiratory and Intensive Care Medicine. His PhD, funded by a Medical Research Council Clinical Research Training Fellowship, builds models to identify phenotypes of acute lung injury from multimodal longitudinal data, with a focus on acute hypoxic respiratory failure and ARDS. He originally trained in molecular and cellular biology at the University of Bath.

Ritam Majumdar

Ritam is a third-year PhD student in the Department of Electrical and Electronic Engineering at Imperial College London, supervised by Sonali Parbhoo. His research is in reinforcement learning, with a focus on off-policy evaluation, inverse reinforcement learning, theoretical guarantees and interpretable RL. Before joining Imperial in April 2024, he graduated from the Indian Institute of Technology Guwahati in 2021 with a major in Chemical Engineering and a minor in Mathematics. From August 2021 he was a researcher at the Tata Research Development and Design Centre (TRDDC) in Pune, India, developing foundation models for partial differential equations.

Nikita Narayanan

Nikita is a third-year PhD student in the Department of Electrical and Electronic Engineering at Imperial College London, supervised by Sonali Parbhoo. Her research focuses on continual learning for foundation models and concept-based representations for healthcare. She leads primary and secondary school STEM sessions as part of Imperial’s outreach programme, helping students reinforce skills in maths and physics as well as inspiring them to see how these apply in the real world.

Elizaveta Sheremetyeva

Elizaveta is a PhD student in the AI4Health Centre for Doctoral Training at Imperial College London. Her project, Developing an Artificial Intelligence agent to optimise decision-making during assisted conception, builds reinforcement learning methods for personalised IVF treatment. She is supervised by Thomas Heinis, Waljit Dhillo, Ali Abbara and Sonali Parbhoo. She is interested in how AI can help people make better decisions in the real world, starting with healthcare. She has co-led Women in Computing at Imperial and previously represented England in international chess.

Haoyang Hong

Haoyang is a PhD student at Imperial College London, supervised by Sonali Parbhoo. His research focuses on causal representation learning and robust machine learning for healthcare, particularly out-of-distribution generalisation and causal approaches to improving model robustness. His research explores both the theory and practical effectiveness of these methods, and the development of reliable machine learning models. Haoyang also has experience in clinical predictive modelling, reinforcement learning and large language models, including medical multi-agent systems, with applications in healthcare and clinical decision-making.

Yuxin Wang

Yuxin is a PhD student in Computer Science at LMU Munich, in the Institute of AI in Management supervised by Stefan Feuerriegel, and is affiliated with the Munich Center for Machine Learning (MCML) and relAI. She is visiting the AI4AI Lab. Her research covers semiparametric and debiased machine learning for treatment effects, causal survival analysis under informative censoring, and causal inference for LLM alignment, including off-policy evaluation and policy learning from preference and observational data, with applications in medicine and economics.

Hannah Liu

Hannah is a visiting researcher in the AI4AI Lab, working on safety in foundation models. She is a PhD student in Design Engineering at Imperial College London, supervised by Pietro Ferraro, where her research develops scalable AI governance frameworks that are model- and industry-agnostic and globally applicable, using technical methods as the backbone of governance to address questions in social science and AI public policy. She holds a Master’s in AI Applications and Innovation from Imperial and a Bachelor’s in Cognitive Science from the University of Pennsylvania.

Hannah is also a Research Engineer in AI Policy & Governance at the Responsible AI Institute, where she helps Fortune 100 companies and NHS-affiliated healthcare professionals build proactive AI governance structures. Her work has been adopted by organisations worldwide, her published framework won two awards at the GenAI Berlin conference, and she sits on an OWASP working group on the Agent Control Standard framework. Outside work, she is a competitive swimmer, beginner scuba diver, and puzzle enthusiast.

Pranav Sharma

Pranav is a second-year Electrical and Information Engineering undergraduate at Imperial College London and a Laidlaw Scholar, conducting research under Dr Sonali Parbhoo. His research currently focuses on advancing Concept Bottleneck Models to develop more interpretable and reliable machine learning models for healthcare.

Beyond his research, Pranav helped lead ArterialAge in developing a wearable device for earlier detection of heart failure, becoming UK champions and placing among the global top 10 out of 687 teams at the 2026 Silicon Labs IoT Challenge.

His broader interests lie in translating advances in machine learning into technologies that can make a meaningful real-world impact.