We intensively use a broad range of OSGeo tooling to put together the original source of truth location data for Australia that, among other things, is the most downloaded dataset provided by the Australian government from data.gov.au.

Our team takes things like SRIDs very seriously and we work hard to ensure our data is very accurate. Additionally we use these great GIS tools to construct many other data use cases such as Australian roads, buildings, trees, indigenous lands, disaster response and climate reporting information. We have encountered some fascinating problems that these tools help us solve.

We're constantly working to be more efficient and to better scale, specifically we have been working on "Standard Operating Environments" that are expected to be fast to deploy, work off the shelf and stay up to date. These are used by a large team of geospatial analysts and engineers. For our relatively mature use cases our development stack uses Python + PostGIS, GDAL, PROJ, Shapely, Fiona, GeoPandas, jupyter and much more. Making a development environment is a well trodden path, this should have been easy right. Right?

This talk will briefly touch on some cool uses we have for these tools; outline how we're working to make our processes better and will discuss some of the challenges we encountered in doing this work, in the hopes that others may avoid them.