At Helios AI, I owned the end-to-end data-science workstream for a global price-forecasting product developed with S&P. I delivered weekly predictions for more than 200 fruit and vegetable commodities, with F1 performance above the industry benchmark.
I designed production Python forecasting workflows combining CNNs, LSTMs, SARIMA and OLS, and built SQL and BigQuery transformations for downstream product use. I also integrated LLMs with quantitative model outputs to produce explainable forecasts and forward-looking market insights.
At Agreena (formerly HummingbirdTech), I built machine-learning and deep-learning models to detect tillage and cover crops from satellite imagery. I worked with Sentinel-1, Sentinel-2 and MODIS data, and contributed data-science requirements to a prototype satellite-imagery pipeline.
Earlier, as Product Manager / Founding Member at Delphi AI, I developed a luxury-fashion trend-forecasting concept using historical sales, climate trends, social-media influence and expert judgement. My work also includes research in quantum technologies and machine learning, alongside hands-on fabrication and construction experience at Unto This Last.
United Kingdom