Now

Local LLM phone handler for small businesses, originally built for clinics. Faster Whisper handles speech-to-text, a local LLaMA/Phi-3 model does intent extraction, and Google Calendar API handles booking. Built through the UWE Enterprise Scholarship; working towards a first pilot. The hard part: making it work reliably on commodity hardware with no internet dependency.

MSc dissertation. Forecasting the total cost of the GB electricity balancing mechanism per half-hour settlement period, using Elexon BMRS data. A statistical baseline (LEAR) against XGBoost and LSTM, evaluated with a rolling-window scheme and Diebold-Mariano tests. Data pipeline and feature table complete; model training next. Deadline: January 2027.

At UWE Bristol. Sem 1 results: Data Management Fundamentals 84 (Distinction), Statistical Inference 70 (Distinction). Currently focused on the dissertation while wrapping up remaining modules. Also serving as student representative for the MSc cohort.

Data engineering, ML engineering, or data science, ideally in energy or fintech. I build things end-to-end - pipelines, models, APIs, and web systems. Open to summer/autumn internships now and graduate roles from 2027. If you have something interesting, reach out.