AI & Machine Learning
I build machine learning systems that work outside a notebook — computer-vision detectors trained on messy field imagery, forecasting models that turn historical signals into forward-looking decisions, and applied research that frames open-ended problems into working solutions. Evaluation loops that tell the truth, and inference APIs that hold up under real traffic.
Computer vision
Object detection and segmentation on aerial, satellite and ground imagery.
Forecasting
Time-series prediction on market and operational data — from UK stock price research to evolving trading signals that blend price history, indicators and news sentiment.
Decision intelligence
Systems that don't just classify but decide — expert verdicts, agentic orchestration and guardrails that guide real-world actions.
Model training
Dataset curation, augmentation, transfer learning and honest evaluation.
Inference APIs
Packaging models behind fast, tunable services with latency budgets.
Applied research
General problem solving via ML research — scoping open-ended questions, prototyping approaches and iterating until the model earns its keep.