// data scientist · MSc Data Science (in progress)

I turn messy data into decisions people can act on.

Machine learning, NLP and applied analytics — with a business background that keeps the question, not just the model, in focus.

See my work GitHub LinkedIn CV
Selected work

Customer Churn Analysis

EDA · segmentation

Exploratory analysis and customer segmentation to help a bank reduce credit-card attrition. Identifies who is churning, why, and which at-risk customers are worth re-engaging — with concrete recommendations for a business manager.

RESULT → Attrited customers contacted service ~3× before leaving; low-income users show highest attrition; early-warning segment identified for retention.
PythonpandasseabornTableau

Cost of Living: What the Headline Hides

deployed app · analytics

An interactive dashboard exploring which UK spending categories most affect household budgets — showing that fastest-rising prices aren't always the ones that hurt households most. Built with ONS inflation data.

RESULT → Live dashboard · 12 spending categories · weighted contribution reveals true budget impact.
PythonStreamlitPlotlyDuckDB

House Price Prediction

deployed app · regression

An interactive app that estimates house prices from three property features — end to end, from data cleaning and feature engineering to a live model you can try in the browser.

RESULT → R²: 0.45 · MAE: 7.13 · Linear Regression (selected for parsimony) · live & usable.
Pythonscikit-learnStreamlitfeature engineering

CIFAR-10 Image Classification

computer vision · ML

Image classification model trained on the CIFAR-10 dataset using a Convolutional Neural Network to classify 32×32 colour images into 10 categories. Demonstrates deep learning workflow and model evaluation.

RESULT → ~80% test accuracy · 50,000 training images · 10 classes: aircraft, automobile, bird, cat, deer, dog, frog, horse, ship, truck.
PythonCNNcomputer visionPyTorch

Currently building: a document Q&A app using retrieval-augmented generation (embeddings → retrieval → LLM). Card coming soon.

About

I came into data science from business and operations — leading work where decisions had to be grounded in what the numbers actually said, not guesswork. That's still how I work: start with the real question, then build the model that answers it.

I'm finishing an MSc in Data Science (Data Science pathway), and I care about work that ships and holds up — clean code, honest evaluation, and results someone can use.

Skills

Languages & data

PythonSQLpandasNumPy

Machine learning

scikit-learnclassificationregressionclusteringmodel evaluation

Build & ship

StreamlitAWSGitPlotly

Building toward

LLMsLangChainRAGAWS MLA-C01