An engine lights up
A supermassive black hole in an active galactic nucleus feeds on surrounding matter. Jets and shocks create an extreme particle accelerator, and a neutrino can be born there.
Work overview
What I work on
A first placeholder page for explaining my research, software work, teaching, and infrastructure projects. The exact wording can be refined later.
A neutrino in seven tiny scenes
Almost nothing gets in the way
The neutrino crosses cosmic distances in a nearly straight line. Magnetic fields do not bend it, and most matter is practically transparent to it.
A rare interaction
Only very occasionally does it meet a nucleus in a dense material cloud, rock, or ice. That single collision is the tiny clue we try to turn into a cosmic message.
A microscopic cascade
The interaction releases secondary particles that form a compact shower. For a short moment, the shower carries an excess charge that moves faster than light can travel in the medium.
A radio pulse escapes
That charge imbalance emits a brief, coherent radio flash. In clear polar ice, the pulse can travel far enough to be measured by buried antennas.
Antennas catch the trace
Each station records a tiny waveform. Timing, amplitude, and polarization let us reconstruct where the signal came from and how energetic the event was.
Software improves the telescope
My work uses simulation, deep learning, and differentiable programming to optimize detector layouts and analysis methods before the next instrument is built.
Astroparticle physics and radio detection
My research focuses on radio-based neutrino detection in ice and on understanding how detector geometries influence the physics reach of an experiment.
Differentiable detector optimization
I develop tools around NuRadioOpt to make end-to-end detector design more systematic, reproducible, and compatible with gradient-based optimization.
Maintenance, teaching, and infrastructure
Alongside research code, I work on simulation tooling, documentation, teaching material, and practical infrastructure for reproducible scientific workflows.
Publications & talks
Selected theses and a recent conference talk pulled from available pages.
Ordinal Classification with Neural Networks in DSEA
Bachelor thesis (2022) on ordinal classification using neural networks; includes a short abstract and supporting materials.
Details →The Impact of Topo-Cluster Splitting on Boosted Object Identification in ATLAS
Master thesis (2025) PDF available; discusses topo-cluster splitting and its influence on boosted object ID in ATLAS.
Details →Differentiable End-to-End Optimization of In-Ice Radio Neutrino Detectors
Conference contribution describing differentiable detector optimization (Indico entry).
Details →More detail later
This page is intentionally broad for now. It can later grow into a structured explanation with figures, examples, publications, and links to concrete projects.