Research and portfolio
Selected research and experimental work that demonstrates how Teesside Bytes approaches complex problems. Projects here are not presented as commercially completed client engagements.
A metadata-led decision-support system for studying ancient writing systems. The research combines symbol recognition and visual similarity with artefact context, provenance and interpretable hypotheses to support expert investigation.
Technologies include Java, Spring Boot, PostgreSQL, Python and machine learning experimentation.
Visit the project repositoryA comparative study using biomechanical spinal measurements to explore how machine learning can distinguish normal and abnormal spinal conditions, including disc herniation and spondylolisthesis.
The work investigated missing-value imputation, feature relationships, class imbalance and dimensionality reduction before comparing classification methods such as Random Forest, XGBoost and LightGBM. SHAP was used to examine the influence of individual features on model predictions.
Academic research and model evaluation only; not a clinically validated diagnostic system.
We can discuss a small proof of concept or a focused software improvement before committing to a larger project.
Get in touch