LLM-Based Agentic AI: Foundations, Systems & Applications – PhD (University Funded) Ref: 5729

Supervisors

Professor Shiqiang Wang

Dr George De Ath

Dr Jawad Fayaz

Large language models (LLMs) can read and write text and code, call tools, and follow instructions. They now allow us to build agents that plan and act over many steps instead of giving a single reply. LLM-based agents is becoming a key part of our everyday life and work, handling multi-step actions in a variety of application domains. However, an important open challenge is making these agents work well in real environments, achieving reliable results, correct use of tools, and predictable cost and speed across heterogeneous infrastructures and devices.

To address this challenge, we are recruiting PhD students to work on this exciting topic of LLM-based agentic AI, with the goal of making AI agents reliable, efficient, and collaborative. The research results are expected to become the core of next generation agentic AI systems. In this PhD programme, you will redefine how the world works, learns, and discovers, turning bold ideas into tools used by millions. You will then become one of the few early experts in practical agentic AI.

Potential Research Themes

· Foundations of LLM-based agents. Theoretical modelling, algorithm design & analysis, and experiments, to make LLM-based agents more reliable and useful in practice.

· Broader LLM enablers for agents. Retrieval-augmented generation (RAG) for up-to-date knowledge and tools, efficient fine tuning for specialization under tight budgets, and closed-loop improvement from user or operator feedback.

· Multi-agent collaboration and coordination. Roles, handoffs, and task decomposition for teams of LLM agents that must cope with partial information and changing conditions.

· Agentic systems and distributed infrastructure. LLM routing and placement across device, edge, and cloud, resource management for heterogeneous models, observability, rollback, and graceful failure modes.

· Applications of agentic AI. Apply the developed agentic AI technologies and systems to real-world problems in areas such as air traffic control, environmental intelligence, sustainability, etc. As a specific example, LLM-based agents can reason over complex spatiotemporal datasets (e.g., climate, hydrological, or environmental sensor data) to support risk forecasting, adaptive planning, and sustainability optimisation.

What You Will Get

· An inclusive and open-minded research environment that bridges theoretical thinking and system building, with opportunities to publish and present at top AI/ML venues and validate ideas on real platforms.

· Mentorship focused on visionary brainstorming, in-depth technical discussions, and collaborative work culture on both theory and systems development.

· Opportunities to interact with other academic and industry partners working on advanced LLM systems.

· Access to computing resources for running LLM-based experiments.

Who You Will Work with

The lead supervisor Professor Shiqiang Wang (https://shiqiang.wang) has recently moved to the University of Exeter after nearly 10 years of industry experience in the United States. He will bring a unique vision, hands-on expertise, and team spirit from his experience in both academic and industrial research. A supervisory team including at least one additional secondary supervisor will be established for each student. The prospective secondary supervisors Dr George De Ath and Dr Jawad Fayaz have significant experience in air traffic control and environmental intelligence, respectively, as well as in the general areas of machine learning, uncertainty quantification, and Bayesian modelling. They will provide complementary expertise to bridge agentic AI with real-world impact.

What We Are Looking from You

· Background in Computer Science or a related field.

· Strong motivation to conduct groundbreaking AI research that make a positive impact in the real-world.

· Excellent communication skills in English, both speaking and writing.

· Proficiency in programming and modern ML tooling (e.g., Python, PyTorch); experience with LLMs (e.g., using Hugging Face libraries) is a plus.

· Ability to reason about complex systems and turn ideas into code and running experiments.

· Curiosity, persistence, and team spirit.

· Good to have but not required: Applicants with prior industry experience (including internship) or experience with open-source contributions are particularly encouraged to apply.

The studentship covers international tuition fees plus stipend for living and will be awarded on the basis of merit. Both UK and international applicants are welcome.

Entry requirements

Applicants for this studentship must have obtained, or be about to obtain, a First or Upper Second Class UK Honours degree, or the equivalent qualifications gained outside the UK, in an appropriate area of Computer Science or closely related fields.

If English is not your first language you will need to meet the English language requirements and provide proof of proficiency. Click here for more information.

How to apply

Apply now

To apply, please click the ‘Apply Now’ button above. In the application process you will be asked to upload several documents:

•            CV (if you are currently studying, please mention when you will graduate; if you have written any open-source code, please include a link, e.g., GitHub profile)

•            Letter of application (outlining your academic interests, prior research experience and reasons for wishing to undertake the project).

•            Transcript(s) giving full details of subjects studied and grades/marks obtained (this should be an interim transcript if you are still studying)

•            Names of two referees familiar with your academic work. You are not required to obtain references yourself. We will request references directly from your referees if you are shortlisted.

•            If you are not a national of a majority English-speaking country, you are suggested to submit evidence of your proficiency in English. This is optional at the application stage but will be required before you can be formally enrolled in the PhD programme.

The closing date for applications is 23:59 UK time on 30 November, 2025.  Interviews will be held virtually in the week commencing 8 December, 2025. 

Start date will be April 2026

All application documents must be submitted in English. If the original document is not in English, certified translated copies must also be provided.

Please quote reference 5729 on your application and in any correspondence about this studentship.

Summary