Postdoctoral Research Associate

University of Liverpool; Liverpool; On-site; GBP 40,000 - 46,000; Full time

Job summary

The University of Liverpool invites applications for a full-time post responsible for day-to-day technical ownership of AIM-SAFE Work Package 5, Task 5.1, including benchmarking and selecting open-weight Small Language Models for clinical NLP tasks such as entity recognition and drug-interaction reasoning.

You will design parameter-efficient fine-tuning pipelines (LoRA, QLoRA, adapters) within a reproducible MLOps framework and apply quantisation for CPU NHS deployment; you will integrate

Responsibilities

• Benchmarking and selecting open-weight Small Language Models for clinical NLP tasks such as entity recognition and drug-interaction reasoning
• Designing parameter-efficient fine-tuning pipelines (LoRA, QLoRA, adapters) within a reproducible MLOps framework
• Applying quantisation (GPTQ, AWQ, GGUF/llama.cpp) for CPU-feasible NHS deployment
• Integrating federated and privacy-preserving learning via the CAFEIN platform at CERN
• Embedding the RUC2 clinical safety governance layer
• Leading Deliverable D5.1
• Acting as technical liaison with KAIST-led orchestration team and Knowledgator-led RAG team
• Publishing in high-impact venues and mentoring junior researchers within Intelligent Systems Research Group, EECS

Job description

Hours of work: Full-time

Tenure: The post is available up to 31 May 2030

AIM-SAFE is a Horizon Europe project building a multi-agent clinical reasoning framework from open-weight, locally deployable AI, in partnership with KAIST, Knowledgator, the Royal Marsden NHS Foundation Trust and CERN.

The University of Liverpool leads Work Package 5, Task 5.1, and this post takes day-to-day technical ownership of it: benchmarking and selecting open-weight Small Language Models for clinical NLP tasks such as entity recognition and drug-interaction reasoning; designing parameter-efficient fine-tuning pipelines (LoRA, QLoRA, adapters) within a reproducible MLOps framework; applying quantisation (GPTQ, AWQ, GGUF/llama.cpp) for CPU-feasible NHS deployment; integrating federated and privacy-preserving learning via the CAFEIN platform at CERN; embedding the RUC2 clinical safety governance layer; and leading Deliverable D5.1.

You will also act as the technical liaison with the KAIST-led multi-agent orchestration team and the Knowledgator-led RAG team, publish in high-impact venues, and mentor junior researchers within the Intelligent Systems Research Group, School of Electrical Engineering, Electronics & Computer Science.

If you are still awaiting your PhD to be awarded you will be appointed at Grade 6, spine point 30. Upon written confirmation that you have been awarded your PhD, your salary will be increased to Grade 7, spine point 31.

Our commitment to Equality, Diversity and Inclusion

We are committed to enhancing a workforce as diverse as our community and particularly encourage applicants who are of minoritised genders and ethnic backgrounds, living with a disability, and/or are members of the LGBTQIA+ community.

Closing Date: 12 November 2026 23:30

£39,906 to £46,049 per annum