From gamma-ray bursts to machine intelligence.
I apply the statistical rigor of high-energy astrophysics to modern machine learning — proper model specification, robust Bayesian inference, and the honest quantification of uncertainty.
Whether I'm analyzing distant cosmic explosions or extracting property intelligence from satellite imagery, the principles never change: specify the model correctly, infer honestly, and embrace uncertainty.
Fiddling with the relativistic emission processes behind gamma-ray bursts was my first passion. Today I bring the same thinking to geospatial computer vision and large-scale machine learning.
Proper statistical inference isn't about MCMC sampling or "Big Data." It's about correct model specification, faithful implementation of statistical concepts, and quantifying what you don't know. Everything else follows from that.
Bringing certainty to property data.
Moody's · Cape Analytics
Munich, Germany — geospatial computer vision for the built environment
I develop geospatial computer-vision systems that analyze property characteristics from aerial and satellite imagery — extracting property intelligence for insurance, real estate, and risk assessment.
focus = ["geospatial CV", "property intelligence", "ML infrastructure"]
data_sources = ["aerial imagery", "satellite data", "property records"]
methods = ["deep learning", "Bayesian inference", "uncertainty quantification"]
mission = "bringing certainty to property data"
A trajectory from cosmos to code.
Tools for inference at cosmic scale.
Lead developer of 3ML — the Multi-Mission Maximum Likelihood framework — plus a constellation of open-source libraries for GRB analysis, population synthesis, and Bayesian modeling.
Most Loved
★- Population-synthesis framework
- Synchrotron from cooling electrons
- Cosmological GRB simulator
- Fokker-Planck solver
Fermi-GBM Suite
☄- Detector response generator
- Geometry & plotting
- Automated localization
- Web-based RSP2 tool
Bayesian & Stats
◎- nazgul9★GRB triangulation (Stan + RFFs)
- IPN time-delay simulations
- PyStan helpers
- 3ML spectral interpolation
Multi-Messenger
⬡- Blazar population synthesis
- Neutrino emission modeling
- whimstan2★WHIM search
- Neutrino drilling
Work in high-energy astrophysics & astrostatistics.
Time-resolved GRB Polarization with POLAR and GBM
Burgess, J. M., Kole, M., et al.
arXiv ↗Automatic Detection of Long-duration Transients in Fermi-GBM
Kunzweiler, F., Biltzinger, B., Greiner, J., Burgess, J. M.
arXiv ↗3ML: The Multi-Mission Maximum Likelihood Framework
Vianello, G., Burgess, J. M., et al.
arXiv ↗50+ publications across high-energy astrophysics and astrostatistics. Full list in my curriculum vitae.
Lectures & invited talks.
Recent invited talks: Spectroscopy of GRBs (Rome, 2021) · 3ML Framework (Berkeley, 2021) · Synchrotron Emission (Nanjing, 2019) · Nazgul GRB Triangulation (Sardinia, 2021)
Fellowships & awards.
- Alexander von Humboldt Fellowship2017
- Royal Swedish Academy of Sciences Scholarship2015
- Oskar Klein Postdoctoral Fellowship2014
- AAS Chambliss Award2012
Handle data with care.
We're good at mathematics and modeling, but the astrophysics community often overlooks the statistical literature. A few resources I return to:
- Dos and don'ts of reduced chi-squared
- Understanding astronomical data with Bayesian methods
- Statistics: Handle with Care — the Likelihood Ratio Test
Off the clock, I play bass. There was a time I shared an article with a band that won a Grammy — it was not my band. The last group I did anything serious with was Oto Benga: a great bunch of dudes trying to say something in a nowhere town.