Luke Andrew
I build computer vision systems from end to end, from curating datasets and training deep learning models through to tuning and deploying them. My projects cover object detection, segmentation and tracking multiple objects at once, mostly in Python with PyTorch and MATLAB.
I enjoy building real-world projects that reach people and make a difference, above all in healthcare. My favourite so far identified neural biomarkers for Parkinson's using explainable artificial intelligence (AI) on electroencephalography (EEG) recordings. I am looking for a computer vision or machine learning engineer role close enough to the finished product to see what it does for the people who use it.
I have a passion for solving problems and optimising systems, making things more efficient, lighter and easier to use. Everything I build is tested, kept under version control and checked automatically with continuous integration and continuous delivery (CI/CD), so I know it will hold up in everyday use.
Outside work I am a devoted sportsman, and I often find myself in leadership roles. The feeling of getting better is addictive, and I enjoy learning and growing, which carries straight into how I work.
Seven things I built, from a clinical EEG study to a robot on a Raspberry Pi.
Project 1 of 7
I implemented explainable AI methods on an EEG foundation model to find which brain signals show whether a Parkinson's patient's medication is working. They sat in a slower range than the one used today.
real output: distance ± error bar, green reliable, amber uncertainProject 2 of 7
From a single drone camera, my system finds and follows people and vehicles, works out how far away each one is with an honest error bar, and runs on a laptop with no GPU.
Median distance error, synthetic flights with exact truth
illustration of the robot, not a photoProject 3 of 7
I built a tracked robot from scratch that patrols on its own and can track intruders. Using live footage, it sends alerts, all running on a Raspberry Pi.
Project 4 of 7
I built a system that finds and follows every person in a video, then made it run almost twice as fast on a normal computer without losing any accuracy.
How many unique tracks are there in total?
tool count_unique_tracks sequence MOT17-04 value 20
20
7 deterministic tools. No number without a tool call.
Project 5 of 7
An AI agent that answers plain-English questions about my tracking results, built so it can't make up numbers.
Project 6 of 7
A computer vision system that classifies, detects and outlines waste on a real conveyor belt, made small enough to run on site.
Project 7 of 7
A search engine that finds the most similar images out of 10,000 from a single photo.
MATLAB2025 – 2026
MSc Computer Vision, Robotics and Machine Learning
University of Surrey
Supervised by Dr Shlomi Haar · reference letter available
2025
MSc Engineering, Pre-Masters
University of Sussex
2021 – 2024
BSc Information Technology, Robotics
Eduvos, South Africa
Golden Key Award
2026 – now
Bartender
Wetherspoon
2021 – 2026
Hockey coach
2021 – 2024
Private tutor, secondary school mathematics and science
Self-employed
If you are looking for a spark of curiosity, energy and growth in your team, and have a role that suits my CV, let me know. Email is the fastest way to reach me.
Guildford, UK · Right to work in the UK, no sponsorship needed