Projects


Large Scale GPU Accelerated Traffic Simulation in Real-Time


This project revolutionises traffic simulation by utilising the Graphics Processing Unit for real-time modeling of a UK city's vehicle population, overcoming computational limitations posed by traditional Central Processing Unit methods.


Developing Agent-Based Models of Criminological Theory


A comprehensive agent-based model of offender behaviours in cities. The model utilises reinforcement learning where, agents organically learn behaviours inline with traditional environmental criminology theory, namely, the rational choice perspective.


A Discrete Event Simulation of the Lloyd's of London Insurance Market - Digital Twin


Working with industry stakeholders for six-months to support the research and development of a digital twin for the specialty insurance market, the objective of the research is to identify the emergent properties of the "Underwriter Cycle".


HADES Asynchronous Discrete-Event Simulation Python Package


HADES (HADES Asynchronous Discrete-Event Simulation) is a small, user friendly framework for creating simulations in python!


Automated Electric Vehicle Charging Optimizer - Zero Emission Vehicles in Emerging Markets Initiative [Fujitsu]


Built a pipeline combining data wrangling, route finding, and modeling to simulate fleet trips, optimizing EV charging locations based on routes and population density. Outputs include wait times, charging speed, costs, and CO2 reduction.


PwC UK Housing Market Agent-Based Model in Python 2022.


Agent-based models built in the late 90's early 2000s were subject to constrained programming frameworks, lacked extensibility, and are exclusively accessible by a small audience. We developed a remake of the famous UK Housing Market Model in Python.


Synthetic Population Synthesis using Large-Language Models [Fujitsu]


A synthetic population generator using empirical UK and EU statistics. The platform leverages mathematical models and large-language APIs to generate and validate spatio-temporal household demographic data across cities for business applications.


Quantifying Electric Energy Demand from Electric Vehicles in Urban Space.


The Urban Traffic Simulator was extended by utilising classical mechanics to quantify energy demand in urban space. Users can now simulate vehicle behaviour in cities and quantify energy demand for policy.


Developing an Agent-Based Model of Vehicle Behaviour in Urban Space.


The Unity 3D software stack was utilised to develop an open-source agent-based model of realistic vehicle activity in an urban street network with granular features such as traction control, drag, mass.


Quantifying the Relationship Between Driver Behaviour in Relation to Speeding and the Number of Vehicles in An Urban Street Network.


A collaborative project with academics and post-graduate researchers looking at the relationship between adherence to speed limits and vehicle density in an urban street network. The project led to the publication of a peer-reviewed article.


Object Detection using Recurrent Neural Networks.


This project involved several colleagues from the School of Geography. The goal was to apply deep learning techniques to drone footage captured in Peruvian forests to quantify the amount of a specific tree that produces fruit.


Applying Statistical Techniques to Sugarcane datasets captured in Brazil.


Building a dataset on N20 sugarcane crop growth in Brazil with applied neural networks. The goal was to identify the optimal configuration of environmental/resource factors that would enhance the growth of Sugarcane while minimising carbon footprint.


Data Visualisations of the relationship between Universities and Local Councils.


A collaboration with a team led by Professor Adam Crawford at the University of Leeds, School of Social Sciences. Produced visualisations of a dataset from surveys to identify relationships between university departments and local councils.


Modelling the Supply and Demand of UK Police Forces, an Agent Based Modelling Approach.


The development of a novel algorithm for an agent based model on the distribution of police resources. The algorithm was embedded in the model to devise the optimal strategy on distributing resources across space.


Incidental Learning of Collocations in an Academic Lecture Through Different Input Modes - Article.


I supported researchers in the School of Education, University of Leeds by providing statistical analysis support for a research project titled: Incidental Learning of Collocations in an Academic Lecture Through Different Input Modes.


Cultivating ‘communities of practice’ to tackle civic policy challenges: insights from local government-academic collaboration in Leeds


Contributed to a project that consisted of data from the University of Leeds and Leeds City Council to establish the relationship between the University and council to try identify civic policy challenges.


Spatial & Demographic Behaviour Selection Model Training Data Curation [Fujitsu]


Utilising cloud-based computer models to de-aggregate macro statistics on origin-destination trips to identify individual-level empirical trips with modes for training our proprietary decision choice machine learning models.