PhD Studentship: Robust, Certified, and Scalable Federated Machine Unlearning for Privacy-Preserving AI

Overview

The Computer Science department at the University of Surrey is offering a fully funded PhD studentship on robust, certified and scalable federated machine unlearning for privacy-preserving AI. Federated learning is increasingly used in sensitive areas such as healthcare and finance, where data cannot be pooled. Yet although GDPR sets out a right to be forgotten, current federated practice offers no robust, scalable or provable way to guarantee that the influence of specific data is removed from a trained model.

The student will develop new federated unlearning algorithms that deliver genuine, verifiable erasure without sacrificing model performance or participant privacy. Research directions include certified unlearning with provable erasure bounds, robustness against adversarial relearning, federated-specific benchmarks and auditing tools, scalable unlearning for large foundation models, and using unlearning to remove harmful knowledge as part of trustworthy AI. The project is supervised by Dr Pedro Porto Buarque de Gusmao and Dr Frank Guerin.

Who can apply

The studentship is open to candidates who pay UK/home rate fees. Applicants need a first-class or strong upper-second-class undergraduate degree, or a Master's degree, in Computer Science, Mathematics, Statistics or a closely related field, together with a solid grounding in machine learning and/or probability and statistics. Python proficiency, familiarity with frameworks such as PyTorch or JAX, strong analytical skills and good written and spoken English are also needed. Experience in federated learning, differential privacy, adversarial robustness, distributed systems or LLM fine-tuning is desirable but not required.

What you get

Funding covers home university fees, additional research training, travel funds and a stipend at the UKRI standard rate of £21,805 for the 2026/27 academic year. Funding is available for 3.5 years. The start date is January 2027, and later starts may be possible by contacting the lead supervisor once the deadline has passed.

How to apply

Applications go through the Computer Science PhD programme. Instead of a research proposal, applicants upload a document giving the project title and the supervisor's name.

Key dates

The application deadline is 31 October 2026.