Projects

A Comparative Study of Retrieval-Augmented Generation (RAG) and Model Conte

RAG vs MCP, AI Retrieval Frameworks used in Finance

SEBASTIAN BITTAR

A Distributed Framework To Enhance Inter-Domain Routing Security

An incrementally deployable overlay management plane framework for enhancing inter-domain routing security through distributed route verification, addressing the inherent lack of validation mechanisms within the Border Gateway Protocol (BGP).

Binita Makanji

A Simulation-Based Autonomous Robot for Construction Site Safety Surveying

An autonomous ROS2-powered robot that surveys construction sites in simulation, detecting and classifying hazards across varied terrain types while mapping their precise locations in real time.

BASHIR ULLAH

A.I.M.S - Autonomous Interior Mapping System

Building affordable rescue robot vision: real-time human detection and 3D mapping using low-cost USB cameras

Balraj Bhuller

Additive Manufacturing Oil and Gas Industry, Mechanical Testing of ER70S-6

Research of Additive Manufacturing using Wire Laser DED Technology for replacing cast iron and steel components, Mechanical Testing and Analysis of ER70S-6.

Michael Roberts

AI Sentiment Analysis with Grounded Summarisation for Pharma Literature

A RAG-based system that summarises antidepressant safety literature using trusted evidence, citations and sentiment analysis to make AI-generated medical information more transparent and reliable.

Siya Asthana

AI-DRIVEN DIGITAL TWIN RADAR SYSTEM FOR DRTRCTION AND CLUTTER MANAGEMENT

Developed an AI-driven Digital Twin radar system integrating CNN and reinforcement learning to improve target detection, reduce false alarms, and effectively manage wind turbine and environmental clutter in dynamic scenarios.

WARREN D COSTA

AI-Powered Smart Wardrobe and Outfit Recommendation System

An AI-powered smart wardrobe system that analyses clothing images and generates personalised outfit recommendations based on style, season, and climate to improve wardrobe efficiency and sustainability.

Himashi Rathnasekara

Al-Powered Steganalysis: Detecting Hidden Data in Digital Images

Al-Powered Steganalysis: Detecting Hidden Data in Digital Images Using Deep Learning A ConvNeXt-Based Hybrid Detection System with Explainable Visual Analysis and Gradio Interface

Guevara Ayos Mahinga

Benchmarking Prompt Injection Attacks Against Large Language Models (LLMs).

This project investigates AI security; whether quantisation affects an LLM’s vulnerability to direct prompt injection attacks, helping identify which model precisions best balance security with business needs.

Aurelia Brzezowska

CFD Simulation of Reactant Transport in Low-Temperature PEM Fuel Cells

How small changes to a fuel cell's internal channels can deliver more power — explored through detailed 3D simulation of reactant flow in PEM fuel cells.

Usman Ahmad

CIA-SecEval: A Simulation Framework for Evaluating Algorithmic Trade-offs i

CIA-SecEval: A Simulation Framework for Evaluating Algorithmic Trade-offs in Multi-Drone Networks

VIKTOR SALIHU

Detecting Anomaly & GPS-Assisted Geofence Bypass Attacks in UAVs

Detecting anomaly and GPS-assisted geofence bypass attacks in UAVs using sensor fusion and LSTM anomaly detection to improve autonomous flight security

Mohamed Afeed Nawsath

Evaluating the Efficacy of Gamified Feedback Systems on Driver Behaviour

A gamified driving feedback system that uses real-time telemetry data to encourage safer driving behaviour through interactive feedback within a simulated environment.

Saira Khan

Flow separation Control using Delta Type and Rectangular Vortex Generators

Preventing aerodynamic stall at high angles of attack using passive flow control devices.

callum baker

Green Hydrogen for Smart Energy Management & Net Zero Mobility

A smart energy management project using green hydrogen storage to reduce CO₂ emissions and improve economic performance through electricity market participation in support of the UK net zero targets.

Sobhan ّFarjam Keleshteri

Hybrid Quantum-Classical Deep Learning for DFU Classification

A hybrid quantum-classical medical AI project for diabetic foot ulcer classification, highlighting why robust evaluation matters.

AVISHKA PRAMUDITHA WARNAKULASURIYA FERNANDO

Implementing Microcontrollers and Sensors to Determine Roller Coaster Track

This project explores the use of microcontrollers and sensors in order to determine emotions of roller coaster riders. This then determines which ride path will be chosen.

Hannah Arnold

InfoGuard: Gamified Phishing Awareness Training Platform

A gamified, browser-based phishing awareness tool that teaches primary and secondary school students to identify online threats through realistic social media DM simulations.

omar bashir

Intelligent SMS Evidence Classification Framework for Mobile Forensics

Explainable AI framework for classifying SMS evidence in mobile phone forensic investigations.

Elaine Fernandes

Lab Generation Tool for the Automated Provisioning of Dynamic Ethical Hacking

Ethical Hacking lab generator tool to provision Docker containers for use in formal assignments and as a learning support mechanism for students studying Cyber Security.

Connor Parkinson

Latency-Aware Predictive Optimisation for Real-Time Edge Based Behaviour

Adaptive edge AI framework using predictive latency-aware scheduling to optimise real-time behavioural inference under dynamic workloads while reducing deadline violations in resource-constrained distributed systems.

Mihran Ali

Latency, Accuracy, and Optimisation for UAV Mapping Systems

A web-based UAV simulation that optimises drone routes, avoids obstacles, and compares A* with Dijkstra using live map telemetry.

BRANDON CASTILLO

MediVoice

MediVoice is an AI-powered GP receptionist system that automates NHS appointment booking using NLP, real-time scheduling, and voice interaction.

JABBAR ABDUL

Model-Assisted Image Annotation with Incrementally Trained Models

Can incremental training be used to create image classification models with minimal labelled data?

J Russ

Music Coach - An Adaptive Software Application for Piano Practice

An adaptive full stack web application utilising Web MIDI API and generative AI to provide music students with real-time, diagnostic feedback and targeted practice recommendation.

Jemima Orakwue

Neuromorphic Context-Aware Sensing for Energy-Efficient Mobile AI

A simulation framework that cuts mobile-AI energy by 80% on UCI HAR by combining context-aware sensor gating with rate-coded spiking neural networks, at under 2pp accuracy cost.

Yusuf Bhula

Optimising Bentley Blower Performance using 1D CFD Simulation Techniques

1D CFD modelling to optimise a 1920s Bentley Blower. Utilising a Ricardo WAVE digital twin, simulation achieved a 16.8% power increase, safeguarding historic integrity through virtual prototyping.

Spencer Potts

Oracle-Based Data Validation for Simulated Digital Twin Environments

A proof-of-concept system evaluating oracle-based data validation for simulated digital twin environments, addressing the critical gap in pre-ingestion data trustworthiness within cyber-physical systems.

Chidube Anike

Real-Time IoT Plant Health Monitoring and Alert System

A real-time environmental monitoring system using an ESP32 core and Adafruit IO to provide qualitative insights into domestic plant care.

Eleanor Weaver

Robotic Arm Project

This project is a Robotic Articulated Arm that is designed to show how force feedback could be applied in a robotic system.

Max McGill

S.O.N.A.R - (Sound-Optimising-Noise-Aware-Robot)

“The Robot that sees with sound and observes with intelligence!”

Rachel Kirkham

SceneAssist - A Lightweight AI Framework

A lightweight AI framework that delivers real-time spoken scene descriptions for visually impaired users using on-device computer vision and contextual scene understanding.

Yahya Ali

Semantic Mapping for Real-time Urban Event Discovery

A web platform that uses machine learning to match city events to your interests in real time, replacing cluttered map pins with intelligent, personalised visual signals.

Sorin Mihalache

Threat Actor Classification and Behavioural Fingerprinting via Deception

An explainable deception-analysis pipeline that turns Deception telemetry into attacker classifications, behavioural fingerprints, and multi-stage attack stories.

Alexander King

Vehicle Speed Estimation using Computer Vision and Deep Learning

A calibration-free computer vision pipeline that estimates vehicle speed from standard traffic camera footage using deep learning and machine learning without any specialist hardware.

Damilola Adedayo

Mellor