01
Spatial optimisation & decision support
When a plan has to balance competing goals across a lot of options.
I build optimisation and scenario models that turn messy multi-objective questions — cost versus emissions versus ecological impact, capacity versus equity — into trade-offs you can actually weigh. Linear and mixed-integer programming (PuLP, HiGHS), Pareto-front analysis, scenario comparison, applied to problems like marine spatial planning, infrastructure siting, and resource allocation. The output is usually a quantified set of options and a way to explore them, not a single "answer."
Related work REEF / North Sea
02
Transit & mobility analysis
When you need to know how people move, and what a change would do.
Accessibility and routing analysis on open transit data — RAPTOR routing, GTFS pipelines, origin–destination and travel-time impact, modal-shift estimation. On the demand side, I work with large GNSS trajectory datasets: cleaning, map-matching, and mode classification that turn raw movement traces into readable patterns. Useful for evaluating a proposed line, mapping who gains and who's left out, or understanding existing flows across a city.
Related work IBX transit routing
03
Remote sensing & environmental monitoring
When the question is about the land itself, over space and time.
Satellite and aerial analysis using open Earth-observation data (Sentinel-1/2, Copernicus) — land cover and change detection, vegetation and tree-canopy mapping, environmental indicators. I span classical raster analysis through ML-based segmentation, and one-off assessments through monitoring pipelines. The environmental-science background means the work is framed around the process, not just the pixels.
Related work Tree-crown detection
04
Geospatial tools & dashboards
When the analysis needs to live somewhere people can use it.
I build the delivery layer, not just the study — interactive maps and dashboards (MapLibre, Leaflet), FastAPI services, and pipelines that take a method from notebook to something a non-technical team can operate. This is often what makes analysis actually get used: a stakeholder-facing tool with the heavy computation precomputed, so exploring scenarios is instant.
Related work District analysis tool