A collection of data science initiatives, machine learning models and programming challenges I've tackled.
An FPL forecasting tool that models player minutes and football events, simulates fixture outcomes, and serves expected points through a cloud-hosted batch API and interactive dashboard.
Building on a first-class dissertation in probabilistic machine-learning for football, I rebuilt a bespoke xG model from the ground up, implementing an interactive front-end plotter to explore and visualise model predictions.
A Premier League table prediction game with private leagues, live scoring and an Elo benchmark that simulates the remaining season 10,000 times. Built with React and Flask, and deployed on Oracle Cloud Infrastructure.
Engine to pick the optimal Fantasy Premier League Challenge squad each gameweek. Adjusts ML point projections for bespoke scoring rules, solves an integer linear programme to maximise expected points, and includes a hindsight pass to compare predicted and actual outcomes.
An ongoing prototype study of multimodal LLM suggestions for augmentative and alternative communication. Early task results are encouraging; the current sample is mostly neurotypical and results are not yet established for the intended AAC user population.
Project focused on predicting theme park crowd levels and ride wait times using historical data, machine learning models and a Streamlit dashboard for interactive exploration.
Web application using the TMDb API for personalised media tracking and watchlists. Engineered a recommendation engine via the Groq Inference API to provide high speed suggestions based on natural language inputs.