Research Scientist
Push the research edge — machine learning on satellite imagery, spatial statistics, and novel geospatial methods.
About Afterrealism
Afterrealism is a small geospatial software consultancy in Canberra. We build maps, data platforms, and cloud systems for organisations across Australia to a standard set inside Geoscience Australia and CSIRO.
Small is the point: you work directly with the founder, your decisions ship, and you can see exactly where your work lands.
About the role
You'll lead the research end of our practice: applying machine learning to Earth observation imagery, evaluating what actually works, and turning promising methods into production capability.
Our founder published research on deep learning for geospatial applications and keeps the company close to the research frontier — segmentation, detection, super-resolution, and foundation models for EO. You'll have room to experiment, publish, and build in the open.
What you'll do
- Design and run ML experiments on satellite and aerial imagery
- Build reproducible training and evaluation pipelines
- Evaluate models rigorously and honestly — including when they don't work
- Hand prototypes over to engineering for production
- Write up findings for clients, the blog, and (where it makes sense) publication
What we're looking for
- PhD, or an equivalent research track record, in machine learning, remote sensing, or geospatial science
- Strong PyTorch and Python
- Sound statistics and experimental design
- Clear scientific writing
Nice to have
- Experience with EO foundation models (Clay, SAM, Prithvi)
- Peer-reviewed publications
- Cloud or HPC training experience
- Contributions to open datasets or open-source tools
Apply — Research Scientist
Replies within a few daysA human reads every application. No screening robots.