Home

I develop vertically integrated inference systems for complex scientific data. My work sits at the intersection of astrophysics, statistics, and machine learning, with a focus on building principled generative models that connect physical source modeling, likelihood construction, hierarchical population inference, and scalable computation into coherent frameworks.

I am currently a group leader at the Max Planck Institute for Gravitational Physics (Albert Einstein Institute, AEI) in Hannover, where I lead a pulsar timing array data analysis group. Our research focuses on inference methods for pulsar timing arrays, Bayesian modeling, gravitational-wave search methodology, sampling methods, Gaussian processes, and pulsar timing models. The aim is to improve how nanohertz gravitational-wave signals are modeled, detected, and interpreted.

More broadly, I am interested in generative inference in science: starting from a model of how data are produced, then building statistical and computational methods that allow us to infer the underlying physics with proper uncertainty quantification. This perspective connects much of my work, from gravitational-wave astronomy to modern machine learning.

I obtained my Master's and Ph.D. in theoretical physics at Leiden University, where I worked on gravitational-wave detection and data analysis for pulsar timing arrays. My Ph.D. research helped lay foundations for the Bayesian PTA searches that are still used today. I later worked at AEI, and then at Caltech and the Jet Propulsion Laboratory as an NSF Einstein Fellow, where I contributed to foundational data analysis methods for the NANOGrav collaboration and the International Pulsar Timing Array.

Between 2016 and 2021, I worked at Microsoft as a senior data scientist on machine-learning problems including language models, text analysis, classification, and reinforcement learning. That period broadened my perspective on scalable modeling, practical machine learning, and modern computational workflows, and it continues to shape how I approach scientific methodology today.

In 2022, I returned to AEI to start a new group on pulsar timing array data analysis methods and gravitational-wave detection. Since then, my research has centered on building the statistical and computational tools needed for the next generation of PTA science.

What I build

My work is centered on the development of methods, tools, and collaborations that make advanced inference usable in real scientific practice. A few recurring themes:

Generative models for scientific data.
I work on modeling the full data-generating process rather than only fitting summary statistics or isolated signals. This includes forward models, structured noise models, hierarchical models, and likelihood-based approaches that preserve interpretability and scientific rigor.

Bayesian and statistical inference.
A central part of my research is the design of robust inference methods: posterior sampling, model comparison, uncertainty quantification, Gaussian-process methods, and efficient algorithms for complex high-dimensional problems.

Pulsar timing array methodology.
I build methods for the analysis of PTA data, especially for gravitational-wave detection and characterization. This includes search methods, noise modeling, timing-model inference, and the statistical foundations needed to make the resulting scientific claims reliable.

Scalable computational workflows.
Good inference requires more than good theory. I am interested in the full chain from model design to implementation: numerical methods, efficient software, reproducible workflows, and computational strategies that allow sophisticated models to be used on real datasets.

Translating modern data science into science.
I care strongly about bringing the best ideas from statistics, machine learning, and scientific computing into domain science in a way that is both technically serious and scientifically meaningful. That includes collaboration across fields, mentoring, and developing shared methodological infrastructure.

If you are interested in collaboration, methodology, or data-intensive scientific inference, feel free to get in touch.