Automation, Control & Robotics · MSc Distinction · United Kingdom
Machine learning that holds up outside the lab.
I build vision, sensing and forecasting systems, then test whether their behaviour survives contact with the real world — noisy inputs, distribution shift, and the people who have to trust the output.
Thirteen peer-reviewed publications, two competitively funded research grants, and eight years teaching engineering.
▍Research focus
My route into AI ran through control theory, sensors and hardware. That order matters: it means I care less about a headline metric than about what a system does on the day the inputs drift.
Do model explanations stay faithful when the data distribution shifts? Attribution methods are widely trusted and rarely stress-tested.
Renewable energy forecasting as the proving ground: predictions that report their own confidence are the ones operators can actually act on.
CNNs, transfer learning and RGB-D perception applied to traffic, agriculture and assistive robotics — trained on data collected in the field, not downloaded.
Systems evaluated with the people who use them. Usability sessions, accessible interfaces, and honest reporting of where the system failed.
▍Selected work
TETFund grant · 2022
A GoogLeNet CNN trained to classify traffic state on the Ondo–Akure highway in Nigeria, a corridor with a heavy crash record and no automated monitoring. The model separates free-flow, congested and crash-indicative conditions from live camera frames so incidents can be flagged without a human watching the feed.
GoogLeNet · Transfer learning · MATLAB / Python · Field-collected video
TETFund grant · 2021
Designed for a campus with unreliable mains power during the pandemic: IR proximity sensors trigger solar-powered DC pumps, so the unit runs entirely off-grid. I ran the user evaluation with students on site, measuring detection reliability and usability in ordinary daily use rather than on a bench.
IR proximity sensing · Solar DC · Microcontroller · Usability study
MSc thesis · Distinction
My MSc dissertation at Sheffield Hallam: a controlled comparison of deep network architectures on a fine-grained flower classification task, holding data, augmentation and training budget fixed so architecture is the only variable. The interesting result is not which network wins but how differently they behave once you look past top-line accuracy.
TensorFlow · PyTorch · CNN architectures · Statistical comparison
ICESET 2022 · Presented
A vision pipeline that lets an autonomous service robot identify and locate common household objects, motivated by assistive use for elderly and disabled people. Objects were modelled with a V4R RGB-D rig and the recogniser integrated into ROS, with a block-based Blockly front end so a carer can configure robot behaviour without writing code.
V4R RGB-D · ROS · Blockly · Point-cloud processing
Five publications
Weather and pollutant monitoring, smart irrigation, gas-leak detection, roadside alcohol screening and a GPS vehicle security unit — built and published between 2021 and 2022. Different sensors, one discipline: get calibrated measurements off cheap hardware, over the network, and in front of someone who can act on them.
ESP-32 · Arduino · GPS · Gas & environmental sensors · Power BI
▍Demo reel · work walkthrough
▍Peer-reviewed record
Thirteen entries across journals, conference proceedings and one thesis, 2017–2024. Corresponding author on three.
▍Research funding
Nigerian Tertiary Education Trust Fund (TETFund) institutional research grant. Covered data collection on the Ondo–Akure highway, model development and dissemination.
TETFund institutional research grant. Funded design, fabrication and the on-campus user evaluation of the contactless unit.
▍Professional experience
▍Education, recognition & membership
Advanced Control Systems · Robotics · Machine Vision · Artificial Intelligence · Industrial Automation.
Recognised for proposing and leading the roll-out of Google Suite for Education across the polytechnic — a deployment valued at over ₦2,000,000, provisioning 2,000 student and staff accounts, with specific commendation for its application to assistive learning.
Commended by the Rector and senior management for performance in a recruitment screening exercise described as well planned and brilliantly executed.
Member, IEEE (98012107) · Member, IET (1101195165) · Registered Engineer, COREN Nigeria (R63246).
Paper presenter, ICESET 2022, Federal Polytechnic Ile-Oluji · Co-author, SEET 2021 Annual Conference, Federal University of Technology Akure.
Python · MATLAB · TensorFlow · PyTorch
SQL · Power BI · Tableau · Statistical analysis · Data visualisation
CNNs · GoogLeNet · Transfer learning · Image classification · Model evaluation
Arduino · ESP-32 · IoT systems · Sensor integration · Microcontrollers
User-centred design · Usability evaluation · Requirements gathering · Accessible interfaces
LaTeX · GitHub · ROS · Google Scholar · ResearchGate
▍Outreach & service
One-to-one mentoring for young people through a grassroots UK charity, plus school-based sessions on the Amplify Me programme — helping students build confidence, find career pathways and meet professionals they would not otherwise have access to.
Workshops and career talks in schools and community groups, showing how AI, robotics and data science look in practice. I speak as a role model from an international and disability background, which changes who students think engineering is for.
Computer setup, system configuration and fault diagnosis for the Goldthorpe branch in South Yorkshire, keeping the charity’s day-to-day digital equipment running. I also sit on the Ijebu Jesa Community IT Committee, managing community sites and IT procurement advice.
▍Open to PhD positions & research roles
I’m looking for doctoral research or an applied research role in trustworthy machine learning, computer vision, or sensor-driven systems. If you have a problem where the model has to be right when it matters, I’d like to hear about it.