Personal Homepage
Welcome to my personal space on the web. I use it to make some information accessible. Services currently provided are:
- blog.pauldebus.de: The blog I sometimes use to report from me and my life.
- snips.pauldebus.de : A small personal project I created to keep track of code snippets I produce while at work or in university and want to find again later.
- github.com/pauldebus: My GitHub account where some of my projects are hosted publicly and you are encouraged to use them and provide feedback.
- git.pauldebus.de/paul: My self-hosted git server where I keep more of my projects and experiments.
- pauldebus.de/publications: An overview over my scientific publications.
About me
I am a photogrammetry and computer vision researcher with more than seven years of applied research in UAS-based imaging, 3D reconstruction, and AI methods - covering the whole pipeline from sensor data acquisition to spatial data interpretation. I did my Dr.-Ing. in Computer Vision at the Bauhaus-University Weimar in Germany, with a dissertation on object-aware UAS flight path planning for image-based 3D structure inspection.
The dissertation grew out of my years as a research assistant at the Chair of Computer Vision in Engineering in Weimar, where I worked on the inspection of infrastructure with pre-planned drone flight paths, taught computer vision courses at Bachelor and Master level, supervised student theses, and acquired industry-funded research projects. I have presented and discussed my work at international conferences and in peer-reviewed venues, and I stay involved in the community as a scientific reviewer (AUTCON, ISPRS, JOSS) and contributor to ISO 7817-3. My research projects include the inspection of dams and concrete structures (Infralytica, Weimar), the evaluation of AI methods for damage detection in concrete (ev.AI.luate), and automated quality control in concrete prefabrication (R2U Technologies, at CERIS / Instituto Superior Técnico in Lisbon).
These days I work as a software engineer at Freiheit GmbH in Lisbon, Portugal, building business-critical web applications and backend systems - bringing the same rigor from research into production software, developer tooling, and CI/CD. Before that, I was a machine learning scientist at Revincus, developing a framework for the detection of events in audio streams. Although I started out as a civil engineer, programming has long been the part I enjoy most, and by now it is simply my profession. I still like writing Python and Julia, but these days I mostly work in TypeScript, JavaScript and Golang.
I care about open source: I co-maintain ImageProjectiveGeometry.jl in the JuliaImages organization and build developer tools in the open, like xsd-to-zod.
You can contact me at paul (at) pauldebus.de to which I normally respond within the day.