SELECTED WORK
The work behind the services.
Models, applications and the engineering decisions that turn an idea into a useful result.
1,022 test rows
Random Forest · Original saved ROC points
Explainable ML for stroke classification.
Data preparation, four classifiers and interpretable results — with the performance trade-offs made visible.
View case studyImage labels → activations → segmentation
From image labels to segmentation.
Data preparation, classifier training and evaluation — with reported segmentation comparisons and real mask examples.
View case studyUseful solutions.
Response variation.Candidates per 5×5 display bin
Real saved archive · Re-binned
Evolving prompts for creative LLM problem solving.
MSc research into how evolutionary search explores prompt styles and their quality–variation trade-offs in LLM problem solving.
View case studyAdvisory agents · Separate execution authority
Coordinated agents. Traceable execution.
A paper-trading application connecting specialist AI analysis, automated strategy checks, risk controls and an inspectable decision graph.
View case studyPersisted tasks → Explained matches → User follow-up
Job search, connected.
A full-stack AI application connecting job discovery, profile-based checks, explained fit scores and application tracking in one workflow.
View case studyLET’S TALK
What would you like to build?
Tell me about your data, your workflow or the idea you want to explore.