Selected engagements
Quantitative modelling and data engineering across staffing, labour markets, climate-litigation research, financial services, automotive SaaS, and public research. Each card opens a fuller writeup.
Time2Fill — vacancy fill forecasting
Daily forecasts of remaining time-to-fill for open vacancies, using survival models that update as the hiring process unfolds.
LabourLens — labour-market dynamics
A market-dynamics model that reads external job-ad flow so delivery can see whether the labour market is tightening or easing.
Climpact — open-weight expert on climate trials
Fine-tuning open-weight models on a golden set of court trials so the project has a specialist reader, not a generic chatbot.
B2B lifetime value & next-best-product recommender
Client-level value estimates with uncertainty, and a recommender for which service to propose next — built so the scores stay valid once they start shaping commercial action.
NLP for unstructured operational text
Language-model pipelines that turn large volumes of unstructured text into structured signal for reporting and search.
Macroeconomic scenario simulation
Constrained optimisation over linked macroeconomic variables, so scenario work can ask what is feasible — not only what happens if.
Distributed ingest pipeline re-architecture
Moved a growing ingest pipeline off single-node processing onto a distributed engine so throughput scales with the data.
Air-pollution mortality Health Impact Assessment
Epidemiological assessment of mortality burden from urban air pollution under alternative traffic scenarios.