Expression of Interest: MSCA Postdoctoral Fellowship University

Positions & Opportunities Postdoctoral Fellowship Expression of Interest: MSCA Postdoctoral Fellowship at the University of Surrey

July 12, 2026

The University of Surrey is seeking an outstanding postdoctoral researcher, with a strong background in computer science, artificial intelligence, computer vision, machine learning, data science, or a closely related field, to co‑develop an application for a Marie Sklodowska‑Curie Actions (MSCA) Postdoctoral Fellowship. The project’s title is "Privacy‑preserving AI for understanding behaviour in real‑world environments".

The selected candidate would work with Surrey’s AI, sustainability, psychology, and hospitality research communities to develop a new privacy‑preserving AI methodology for understanding human behaviour in real‑world service environments, sitting at the intersection of computer vision, behavioural sequence modelling, responsible AI, and sustainability. MSCA Postdoctoral Fellowships are prestigious European fellowships that support excellent researchers through international mobility, advanced training, and ambitious research.

A European Postdoctoral Fellowship normally funds a researcher for 12 to 24 months at a host in Europe or a Horizon Europe Associated Country, with the fellow designing the project jointly with the host supervisors and, if funded, receiving a full‑time research position, training, supervision, and career‑development support.

Much research on human behaviour relies on surveys, interviews, short laboratory tasks, or isolated observations. These methods are valuable, but they often miss how behaviour unfolds across time, space, and multiple decision points, which is especially limiting for sustainability‑related behaviour in real environments, where outcomes are shaped by sequences of movement, interaction, hesitation, repetition, and response to the surroundings. The proposed fellowship would develop an early‑stage AI research tool that converts real‑world observational signals into anonymised behavioural metrics.

The aim is not to identify individuals, classify personal characteristics, or store intrusive data, but to explore how AI can support privacy‑preserving analysis of behavioural patterns in bounded service environments. The first application would be developed in a sustainability‑relevant hospitality setting, where the University has already secured access to a large operational field site, offering an unusual chance to move beyond laboratory proof of concept toward a validated, ethically robust prototype.

Working with the Surrey team, the fellow would take the prototype from proof of concept toward a validated research tool.

• Design and build a minimum viable prototype for privacy‑preserving behavioural AI
• Develop or adapt computer vision and machine learning methods for detecting non‑identifying behavioural events
• Convert detected events into behavioural sequences interpretable by social scientists
• Develop validation protocols comparing AI outputs with human‑coded observations
• Work with behavioural scientists to ensure outputs are meaningful, not just computationally detectable
• Contribute to ethics, privacy‑by‑design, and responsible AI procedures
• Test the prototype in controlled settings and, where feasible, a real‑world hospitality field site
• Publish interdisciplinary research and prepare follow‑on funding applications

The target is a validated prototype suitable for further research funding and future deployment studies, not a finished commercial product.

Surrey is particularly interested in candidates holding a PhD, or expecting one by the MSCA call deadline, in areas such as computer vision, machine learning or AI, human activity recognition, multimodal sensing, signal processing, data science, robotics perception, privacy‑preserving AI, human‑centred or responsible AI, or computational social science. Essential or highly desirable skills include strong programming ability, preferably in Python; experience with machine learning or deep learning frameworks;

a grounding in computer vision, object detection, tracking, or sequence modelling; the ability to…