Projects

Projects

Selected software, modelling, automation, and prototyping work, organised into personal and professional projects.

Professional Projects

Applied work delivered in research and industry settings.

Selected projects completed in academic and professional environments, focused on uncertainty analysis, automation, and practical delivery.

Document processing automation at SCIRIS.

An automated document pipeline built to retrieve files, extract text, identify context with LLM support, and update structured records with minimal manual handling.

Problem

Reduce manual extraction and database entry work.

Role

Designed the end-to-end ingestion and update workflow.

Tools

Python, pandas, Azure Document Intelligence, Azure OpenAI.

Result

About 37 seconds per project with more than 80% effort reduction.

Monte Carlo simulation of X-ray photon transport.

A stochastic simulation engine for photon generation and photon-matter interactions, including material absorption, attenuation data, and statistical error behaviour in Monte Carlo runs.

Problem

Model X-ray interactions with matter using stochastic sampling.

Role

Designed and implemented the simulation workflow.

Tools

Python, acceptance-rejection sampling, attenuation data.

Output

Reusable modelling framework for photon transport studies.

Particle motion near black holes.

A project on relativistic particle trajectories in strong-gravity regimes, combining derived geodesic equations with computational exploration of Schwarzschild and Kerr backgrounds.

Problem

Analyse particle motion across physically distinct relativistic regimes.

Role

Derived equations and implemented the analysis tools.

Tools

Python, relativity, ODE/PDE modelling.

Output

Research-led computational study of compact-object motion.

Satellite radar altimetry uncertainty analysis platform.

Interactive Python-based work to document, explore, and report end-to-end uncertainty propagation in satellite radar altimetry re-tracking.

Problem

Make complex measurement chains and uncertainty sources transparent.

Role

Built reusable analysis and reporting workflows.

Tools

Python, Jupyter, Django, pandas, SciPy, diagrams, equations.

Output

Documented framework for uncertainty modelling and communication.

Personal Projects

Independent work across science, data, and prototyping.

Self-directed projects spanning scientific computing, forecasting, machine learning, embedded builds, and mechanical prototyping.

UK greenhouse gas emissions trend and forecasting.

A reproducible pipeline to clean the ONS atmospheric emissions dataset, visualise long-run trends, and forecast five years ahead with a holdout comparison across baseline, ETS, and ARIMA models.

Problem

Build a transparent forecasting and comparison workflow.

Role

Built the full pipeline as a personal project.

Tools

Python, pandas, ETS, ARIMA, visualisation, backtesting.

Output

Clean long-run dataset, benchmarked models, and forecast comparison.

Traditional Japanese blacksmithing and tool-making.

Practical work in traditional Japanese blacksmithing, using tamahagane from the tatara process and refining it through repeated forge-welding before shaping, heat-treating, and finishing functional tools.

Problem

Take raw smelted steel through refinement, shaping, hardening, and finishing.

Role

Managed the full process from steel preparation to completed hand-forged tools.

Tools

Tamahagane, forge-welding, yaki-ire, hamon development, traditional finishing.

Output

Custom tools produced through traditional steelmaking and differential hardening methods.

Portable Linux computer with custom CAD enclosure.

A Raspberry Pi-based portable computer designed for running Python, with a full 3D CAD enclosure covering screen, keyboard, battery pack, cable routing, ventilation, and fit tolerance.

Problem

Integrate hardware, enclosure design, and usability in one build.

Role

Designed, built, printed, tested, and iterated the full system.

Tools

Raspberry Pi, 3D CAD, FDM printing, thermal checks.

Output

Portable Linux-based device treated as a complete integrated system.

Perceptron-based neural models for classification.

A school project implementing a perceptron-based neural model to solve mathematical classification problems and introduce optimisation-driven modelling.

Problem

Build a simple trainable model for classification tasks.

Role

Implemented the model and explored parameter tuning.

Tools

C++, perceptron logic, numerical optimisation.

Output

Early AI project linking computation with mathematical modelling.

3D printers and mechanical prototyping.

Long-term practical work with FDM 3D printers, print settings, prototype housings, brackets, mounts, and rapid print-test-redesign cycles for mechanical refinement.

Problem

Move quickly from CAD models to reliable physical parts.

Role

Built printers, diagnosed print issues, and refined part designs.

Tools

Cura, AutoCAD, Fusion 360, print tuning, mechanical testing.

Output

Practical prototyping workflow shaped by repeated iteration and testing.