The project will explore a flexible set of research directions, allowing the successful student to shape the work around their interests. Possible topics include sparsity and pruning, where unnecessary model parameters or computations are removed; low-precision and quantised training; efficient use of the key-value cache during generation; and speculative decoding, where a smaller model helps a larger model generate text more quickly. The research may span both algorithm design and practical systems work, including experiments on modern GPUs and open-source language models.
A central aim will be to understand when efficiency methods genuinely reduce training or inference cost, rather than only reducing the theoretical size of a model. The student will develop and evaluate methods using measures such as model quality, training time, generation speed, memory use, energy consumption and hardware utilisation. Depending on the direction taken, the work may also consider efficient fine-tuning, hardware-aware optimisation, or combining several techniques into an end-to-end system.
The project is jointly supported by the University of Bath and NetFM UK Limited, a UK software company interested in secure, practical and cost-effective AI-enabled services. The student will benefit from academic supervision in efficient machine learning and opportunities to discuss real-world deployment constraints with the industry partner. No proprietary company data is required for the core research, so the project can be conducted using open models, datasets and benchmarking tools.
This project would suit a curious and motivated student who enjoys machine learning, experimentation and building efficient software. It offers the opportunity to contribute to an important and fast-moving research area, publish at leading international venues, and develop skills relevant to both academic and industrial careers.
Project Keywords: Large language models; efficient machine learning; sparsity; pruning; quantisation; low-precision training; speculative decoding; KV cache; GPU optimisation; sustainable AI
Candidate Requirements
Applicants should hold, or expect to receive, a First Class or high Upper Second Class UK Honours degree (or equivalent) in Computer Science, Artificial Intelligence, Mathematics, Electronic Engineering or a closely related subject. A master's qualification is advantageous but not essential. Strong programming skills, preferably in Python and PyTorch, and strong understanding of machine learning are highly desirable. Prior experience with large language models, GPU programming, model compression or computer systems would be useful, but is not required.
Enquiries and Applications
Informal enquiries are encouraged and should be directed to Dr Rohit Babbar.
Formal applications should be submitted via the University of Bath’s online application form for a PhD in Computer Science prior to the closing date of this advert.
IMPORTANT:
When completing the application form:
- In the Funding your studies section, select ‘University of Bath URSA’ as the studentship for which you are applying.
- In the Your PhD project section, quote the project title of this project and the name of the lead supervisor in the appropriate boxes.
Failure to complete these two steps will cause delays in processing your application and may cause you to miss the deadline.
More information about applying for a PhD at Bath may be found on our website.
PLEASE BE AWARE: Applications for this project may close earlier than the advertised deadline if a suitable candidate is found. We therefore recommend that you contact the lead supervisor prior to applying and submit your formal application as early as possible.
Funding Notes
Candidates may be considered for a University of Bath studentship tenable for 3.5 years. Funding covers tuition fees, a stipend (£21805 per annum 2026/7) and access to a training support budget.
