I develop and maintain several open-source software packages for cosmological inference and gravitational wave analysis. All code is available on GitHub.
Research Software
Bayesian Optimisation for Cosmology
A machine-learning framework based on Bayesian Optimisation using Gaussian Process regression for
accelerated cosmological inference. This tool enables parameter estimation and model comparison with
10-100 times fewer likelihood evaluations compared to traditional methods. Features include:
- Gaussian Process surrogate modeling of likelihoods
- Efficient acquisition strategies for optimal sampling
- Integration with cosmological analysis pipelines
Comparison of Bayesian Optimisation (BOBE) with traditional MCMC methods. BOBE requires 10-100 times fewer likelihood evaluations while maintaining accuracy in parameter estimation and evidence computations.
Non-parametric Reconstruction Tools
Python toolkit for non-parametric Bayesian reconstruction of primordial curvature power spectra
and equation of state evolution from scalar induced gravitational waves. Designed for analysis
of data from NANOGrav, LISA, and Einstein Telescope.
Non-parametric Bayesian reconstruction of the primordial power spectrum from scalar induced gravitational wave signals, demonstrating the method's ability to recover primordial spectra without assuming specific functional forms.
Technical Skills
Programming & Tools
- Languages: Python, Fortran, Mathematica
- Cosmology Tools: CAMB, Cobaya, enterprise
- Machine Learning: Gaussian Processes, Neural Networks, Symbolic Regression
- Computing: High Performance Computing (HPC), parallel computing