Available for freelance work · open to salaried roles
karstlab · Data science & geospatial
Data in the service of the common good
The studio of Thomas Goossens, data scientist and geographer. My craft: making data intelligible and turning it into reliable, reproducible tools that inform decisions on socially and environmentally driven projects. Rooted in the Earth and environmental sciences, with an ethical stance and open-source tools.
Domains: climate · water · energy · agrometeorology · biodiversity · territory … and beyond
Geospatial
GIS, cartography and spatial data, from the PostGIS database to the web map.
Reproducible data science
R development and packages, from full pipeline to application. Versioned, containerised, reproducible code.
Earth & environment
Domain expertise rooted in the Earth and environmental sciences.
Services
Concretely, how I can help you.
Make your data reliable
Ingesting heterogeneous sources, cleaning, industrialising your data flows, with end-to-end traceability. From scattered data to a clean, versioned and auditable foundation.
Deliverables: reproducible pipeline · data model · documentation
Build the R tool you’re missing
From analysis script to a documented, tested R package, up to the interactive app. A tool your team can re-run, maintain and grow.
Deliverables: R package · Shiny app · reproducible report
Map to decide
Spatial analysis and turning territory into data, from the PostGIS database to the interactive web map. Turning a geographic question into a decision-support tool.
Deliverables: GIS analysis · web map · territorial dashboard
Make your work reproducible
Git, renv, Docker, Quarto: practices and tools for work that can be re-run, audited and handed over. Audit of the existing setup and team upskilling.
Deliverables: reproducible environment · automated reporting · skills transfer
Selected work
A few representative projects.
agrometeoR
Geostatistics module and R package for agrometeorological forecasting.
R · mlr · gstat · sf · open-source package
Regional energy balances
Reproducible industrialisation of the regional energy balances reported to the European Commission.
R · KNIME · PostGIS · Quarto · renv · Git
Nation-wide solar scoring
Geospatial prospecting tool spotting large industrial rooftops for solar panel installation, at the scale of Belgium.
R · sf · osmdata · tidyverse · leaflet · phacochR
Flexibility simulation & energy-data ingestion
Tools simulating the gains of electrical flexibility control, and a hardware-agnostic sensor-data ingestion system.
R · Shiny · Python · Streamlit · TimescaleDB · Elia / ENTSO-E APIs
Mapping new-build real estate prices
Predictive mapping, by statistical sector, of new-build price per m² in Brussels, for the urban charges framework.
R · sf · tidyverse · mlr · ggplot2 · leaflet
Mapping a MOOC's participants
Registration form and interactive map of the participants in L'Archipel du Vivant's Biorégions MOOC, 100% open source and European.
Framaforms · R · tidyverse · Geoapify · uMap
kmZen: mileage allowance tracking
A local-first, privacy-respecting tool for logging mileage allowances in a health centre, currently being turned into a European, GDPR-compliant SaaS.
Nuxt · Vue · TypeScript · Nhost · PostgreSQL · GDPR
Who I work with
Public institutions, research teams, associations, transition and environmental actors: organisations whose mission makes sense.
Why “karstlab”?
A karst is a limestone landscape carved with hidden underground rivers and networks: beneath the surface, everything is connected. That’s my metaphor for data: the hidden infrastructure good decisions rest on. And lab, for scientific rigour and a taste for experimenting. (And honestly, I just love scrambling around these landscapes riddled with caves and springs.)