Researcher in Large Language Models

Oxford Source: WhatJobs

See advert for pay

Fixed-term (36 months / 3 years)

The post holder is a member of the research team for the God and the Machine project, with responsibility for the technical aspects of the project’s experimental work with large language models and other AI tools.

God and the Machine: Frontier Interactions between AI Science and Theology is a research project in the Ian Ramsey Centre for Science and Religion, Faculty of Theology and Religion, University of Oxford, funded by the John Templeton Foundation. The project develops a theological framework for understanding artificial intelligence as a mediating tool, working across two interrelated workstreams: Workstream 1 (Interpretation), which uses large language models as a computational hermeneutic tool, and Workstream 2 (Influence), which develops a theology of AI’s mediation of human formation. The post holder will primarily support Workstream 1 in its technical aspects, undertaking experimental work with large language models and other AI tools and techniques, and will also contribute to Workstream 2 as a technical expert.

The gross annual salary has been set at Point 31 of the National Pay Spine, as appropriate for the career stage and experience of the post, and increments one scale point each October, in line with University of Oxford policy.

The project runs from 1 December 2026 to 30 November 2029; the post is August 2026 2 available for up to 36 months within this period, with the exact start date to be agreed with the successful candidate.

About You

You will hold a PhD/DPhil in Computer Science, Machine Learning, or a closely related field, with a keen interest in theological research on artificial intelligence. You will be able to demonstrate expertise in large language models, including experience deploying and experimenting with local models (for example, interpretability, activation steering, or machine unlearning), as well as be proficiency in Python and relevant machine-learning frameworks.