• Calibration of stellar-parameter estimation from astrophysical spectra:
  Most algorithms employ synthetic stellar spectra. Unfortunately, there is always a discrepancy between synthetic and real spectra, which gives rise systematic misestimations. My task is to assess this bias and to correct for it.
• Galaxy morphologies:
  The visual appearance of galaxies provides a direct observable that correlates tightly with other more astrophysical properties. However, classification of galaxy morphologies is hampered by the rich diversity of shapes, which complicates a decent parametrisation.
• Estimation of the stellar initial mass function and its proclaimed universality:
  IMF estimates are usually based on binned data and neglect errors in stellar-mass estimates. I am developing a method that overcomes both problems. This could help to estimate the IMF of systems as small as young stellar groups and therefore provide serious tests of the proclaimed universality of the IMF.
• Detection of exoplanets in radial-velocity data:
  Most exoplanets are detected using radial-velocity measurements of their host star. Often, inappropriate methods are used to estimate the number of exoplanets, namely reduced chi-squared or p-values (e.g. from FAPs or KS-tests). This is a rare situation where already minorly inappropriate methodology can lead to drastically incorrect results.

• statistical methodology in astronomy:
  • comparison of population proportions, e.g., comparing galaxy-merger rates in active and inactive galaxies
  • supervised classification algorithms (Bayes' classifier, linear/quadratic discriminant analysis, nearest neighbour, tree-augmented naive Bayes, support-vector machines, (boosted) decision trees, random forests)
  • unsupervised classification algorithms, i.e., class discovery (K-means, spectral clustering, kernel PCA, self-organising maps, Gaussian mixture models, bipartite-graph models)
  • dimensionality-reduction techniques (principal component analysis, independent component analysis, non-negative matrix factorisation)
  • regression methods (maximum-likelihood parameter estimation, error estimation, Monte-Carlo methods, basic knowledge of nested sampling algorithms)
  • model assessment (Bayes factors, bias-variance trade-off, reduced chi-squared, Bayes'/Akaike's information criterion, cross validation, bootstrapping)
• programming skills:
• Java: advanced knowledge, e.g., object-oriented programming, performant coding, JUnit testing.
• C++: advanced knowledge, e.g., object-oriented programming, performant coding, using external libraries such as the GNU Scientific Library (methods for root finding, minimisation, numerical integration, etc.) and the ATLAS package (FORTRAN code for linear algebra).
• Python: good knowledge, in particular concerning matplotlib (plotting with python).
• Mathematica: good knowledge, including programming intermediately complex routines.
• HTML: rudimentary knowledge, as exhibited by this homepage.

In April 2003, I started my undergraduate studies in physics at the Free University of Berlin, Germany.

From September 2005 to June 2006, I was an ERASMUS exchange student at the University of Manchester, UK. The curriculum at Manchester was comprised of numerous smaller courses, which offered a great opportunity to try various different fields of physics and to make up my mind what I was really interested in.

Afterwards, I did not return to Berlin, where I could not specialise in astronomy. Instead, I moved to the University of Heidelberg, Germany, where I finished my undergraduate studies with a diploma in physics in April 2009.

In June 2009, I started my PhD at the Max-Planck-Institute for Astronomy, Heidelberg, under the supervision of Knud Jahnke. During the whole time of my PhD, I received a scholarship from the Klaus-Tschira Foundation. In October 2011, I finished my PhD with the degree of magna cum laude.

Since October 2011, I am a postdoc in the Gaia group of Coryn Bailer-Jones at MPIA in Heidelberg.