Postdoctoral Research Associate in Machine Learning applied to Neuroimaging in London

The appointee will join the School of Biomedical Engineering & Imaging Sciences.

The research associate will lead the development of cutting-edge multi-modal MRI foundation models. These models will leverage both imaging data and corresponding radiology reports during training to build comprehensive representations that capture the rich, complementary information contained in medical images and clinical text.

The primary focus of this role is to develop foundation models that can be applied downstream to clinical triaging tasks—helping prioritise cases based on MRI imaging data and associated textual information. By integrating visual and language modalities, these models aim to improve the speed, accuracy, and efficiency of interpreting complex MRI scans, ultimately supporting better patient outcomes.

The successful candidate will lead the development of multi-modal MRI foundation models that integrate imaging data and radiology reports. Using advanced deep learning techniques—including vision-language architectures (e.g., CLIP, BLIP), fine-tuning large language models for clinical NLP, and self-supervised contrastive learning—the models will learn to effectively combine visual and textual information.

By developing these foundation models, you will enable downstream clinical applications focused on triaging adult brain MRI scans—helping healthcare professionals prioritise and interpret MRI scans more efficiently, ultimately improving diagnostic workflows and patient care.

This position offers a unique opportunity to drive innovation at the intersection of AI and medical imaging, making a tangible impact on clinical decision-making and healthcare delivery. This is a full-time post (35 hours per week), and you will be offered a fixed term contract ideally starting from 2nd January 2026 until 1st Jan 2029.

We are seeking candidates with expertise in multi-modal deep learning to support the development of MRI foundation models that integrate imaging data and radiology reports for downstream clinical applications.

• PhD qualified in relevant subject area (or pending results/near completion)
• Experience applying multi-modal models specifically in medical or clinical domains.
• Strong knowledge of MRI data formats (DICOM, NIfTI) and image preprocessing tools (e.g., MONAI, SimpleITK).
• Excellent programming skills, demonstrated through available code or projects, with proficiency in Python and deep learning frameworks like PyTorch, Hugging Face, sklearn, tensorflow.
• Excellent verbal and written communication skills
• Experience with GPU training and handling large medical datasets e.g., large magnetic resonance (neuro)imaging datasets.
• Basic understanding of radiology clinical workflows and radiology report structure.
• The ability to take individual responsibility for planning and undertaking own work, according to clinical and scientific deadlines
• Presenting scientific research in the form of papers, posters or oral presentations
• Understanding of the concepts and application of research ethics
• Experience with the use of computing servers

• Experience fine-tuning large language models (e.g., BERT, BioGPT, MedPaLM) for clinical NLP tasks.
• Experience with cloud or distributed computing environments.
• Familiarity with self-supervised and contrastive learning techniques for aligning text and images (e.g., CLIP, SimCLR).
• Clinical experience, e.g., interaction with clinicians and/or handling of patients
• Familiarity with MLOps tools such as MLflow or Weights & Biases for experiment tracking.