About Me
Samuel Farrens
Cosmologist & Python enthusiast
Research Director · CEA Paris-Saclay
Bio
Sam Farrens is a Research Director in the CosmoStat team of the Astrophysics Department (AIM) at CEA Paris-Saclay. His research focuses on cosmology, machine learning, signal processing, and scientific software development. He has been a member of the Euclid Consortium since 2012, holding Builder status, and contributes to various aspects of the Science Ground Segment, including co-leading OU-LE3 and the 3×2pt pipeline group. He is a strong advocate for open science and actively supports early-career researchers. He holds a PhD in Astrophysics from University College London and an HDR from Université Paris-Saclay.
Where I work
I have been very fortunate to end up working in a very diverse and exciting team called CosmoStat. CosmoStat is a lab based in the astrophysics department at CEA Paris-Saclay, just south-east of Paris.
The team is made up of researchers from astrophysics and signal processing backgrounds, working together to tackle the challenges in cosmology by applying the latest developments in statistics and data science to technical problems like measuring galaxy shapes or modelling the point spread function of an instrument in space.
What I do
Weak Gravitational Lensing
I work on weak gravitational lensing science and pipeline development, with a focus on PSF modelling, shape measurement, and end-to-end data processing. I co-lead OU-LE3 — the Euclid Consortium Science Ground Segment pipeline responsible for producing weak lensing and galaxy clustering data products for the ESA Euclid mission — and contribute to the UNIONS wide-field optical survey.
Machine Learning & AI
I work on applying machine learning and statistical signal processing to problems in observational cosmology, including image reconstruction, source detection, deblending, and the modelling of instrumental effects — primarily in the context of Euclid.
Scientific Software
I have a strong interest in building reliable, well-documented, open-source scientific software. I have led and contributed to several Python packages aimed at solving inverse problems and image reconstruction challenges across astrophysics and other imaging domains.
Open Science
I am a strong advocate for open and reproducible science. I regularly contribute to community training, give tutorials at summer schools and conferences, and actively support early-career researchers in developing good software and data practices.
What I used to do
Prior to joining the CosmoStat team my work was primarily focused on the optical detection and analysis of clusters of galaxies using photometric redshifts. During my PhD I developed a prototype friends-of-friends optical cluster detection algorithm that was further developed and optimised during postdoc positions in Barcelona and Trieste. I also developed several metrics and analysis codes designed to compare the relative performance of various cluster detection codes on Euclid mock galaxy simulations.
In the past I also spent time at NeuroSpin, a world-leading biomedical imaging institute, where I worked with a team aiming to improve the acquisition and reconstruction of brain images using Magnetic Resonance Imaging (MRI).
What I like
If you are curious to find out about some of my non-academic pursuits have a look at my blog where I talk about other things that interest me.