My research sits at the intersection of social economics and computer science, combining agent-based modelling (ABM) with artificial intelligence (AI) to study how personal preferences scale up into wider social and spatial outcomes. The work builds progressively: first by establishing a methodological foundation for ABM within the social sciences, then by applying it to labour markets, and finally by extending it into questions of urban sustainability and design.
The initial stage focuses on labour market dynamics, where ABM is used to examine how preferences for working from home interact with firm recruitment and wage strategies. The modelling shows that once structural conditions such as wages are effectively fixed by market forces, it is individual choices — whether to work remotely, to accept lower wages, or to prioritise flexibility — that decisively shape recruitment outcomes. This part of the research demonstrates how small-scale behavioural heterogeneity can generate system-level effects in employment and productivity.
The ongoing stage broadens this lens to urban environments, asking how community preferences can be incorporated into the planning and siting of green infrastructure such as street trees, pocket parks, and green roofs. Here, the research integrates AI and ABM:
Convolutional neural networks (CNNs) act as surrogates for slow climate models, predicting local cooling from interventions.
Bayesian discrete choice models estimate how residents with different backgrounds value features such as safety, shade, and accessibility.
Optimisation and reinforcement learning methods explore siting strategies that balance cooling, equity, and budgetary constraints.
These AI outputs are then linked with ABM simulations of residents, planners, and maintenance agents, creating a dynamic testbed where optimised strategies are tested against the realities of adoption, capacity, and social legitimacy.
Looking forward, the research is not only technical but also visionary in its implications for planning and design. By embedding community preferences into computational models, it demonstrates pathways for planners and policymakers to move beyond one-size-fits-all solutions. The ambition is to support the creation of cities that are cooler in the face of climate change, fairer in terms of equity, and more responsive to the lived experience of their communities. In this way, the research aims to contribute both to academic debates and to the practical reimagining of how cities are designed and sustained.
