Postdoctoral position in differentiable modelling and cosmological inference for Rubin/LSST (M/F)

CNRS - National Center for Scientific Research; France; On-site; EUR 32,000 - 46,000; Full time

Benefits offered by this job

Health cover

Holiday entitlement

Subsidised meals

Transport discounts

French language courses

Job summary

CNRS - Astroparticule et Cosmologie in Paris invites applications for a two-year postdoctoral position starting December 2026. The project, AstroDeep, aims to develop a unified, probabilistic and differentiable framework for cosmological analysis using forward modelling on LSST data and precursor surveys.

The successful candidate will advance differentiable cosmological models, link parameters to observables, and validate methods within a collaborative, international setting with access to

Qualifications

• PhD, or near completion, in astrophysics, cosmology, physics, statistics or a related field.
• Strong ability to perform cosmological simulations and model large-scale structure evolution.
• Experience with gradient-based inference and Bayesian methods is advantageous.

Responsibilities

• Contribute to the AstroDeep project by developing cosmological inference methods based on forward modelling for LSST analysis.
• Develop and validate differentiable cosmological models for the evolution of large-scale structures and shear observables.
• Create inference workflows to quantify biases, uncertainties and robustness to systematics, including gradients.
• Integrate tools into LSST pipelines, document open-source code and disseminate results via papers and talks.

Skills

Cosmological simulations

Large-scale structure evolution

Weak gravitational shear

Differentiable modelling

Bayesian inference

Gradient-based inference methods

Education

PhD in astrophysics or related field

Tools

JAX

PyTorch

Bayesian sampling

High-performance computing

Job description

Organisation/Company CNRS Department Astroparticule et Cosmologie Research Field Physics Researcher Profile First Stage Researcher (R1) Application Deadline 29 Oct 2026 - 23:59 (UTC) Country France Type of Contract Temporary Job Status Full-time Hours Per Week 35 Offer Starting Date 1 Dec 2026 Is the job funded through the EU Research Framework Programme? Not funded by a EU programme Is the Job related to staff position within a Research Infrastructure? No

Offer Description

• To contribute to the AstroDeep project, jointly led by the APC (Université Paris Cité/CNRS) and CosmoStat at CEA Paris-Saclay, with a view to developing cosmological inference methods based on forward modelling for the analysis of the LSST survey from the Vera C. Rubin Observatory.
• To exploit the cosmological information contained in the non-linear structure of matter fields and weak gravitational shear, beyond traditional statistics, by linking cosmological parameters to observables via physical models and automatically differentiable simulations.

• To develop and validate differentiable cosmological models for the evolution of large-scale structures and gravitational shear observables, whilst assessing their accuracy and performance.
• To develop inference methods that utilise these models and their gradients, to characterise biases and uncertainties, and to study scaling, the marginalisation of nuisance parameters and robustness to systematic effects.
• Apply the workflow to realistic simulations and data from precursor surveys such as DES, then prepare for and, subject to data availability, carry out the cosmological analysis of LSST Y1 within the LSST DESC collaboration.
• Integrate the tools into the LSST pipelines and validate them on volumes representative of the survey; document and distribute the open-source code, and disseminate the results through scientific publications and presentations.

The AstroDeep project aims to harness the full informational content of large Phase IV imaging surveys by developing a unified, probabilistic and differentiable framework for cosmological analysis.
A two-year postdoctoral position is available at the APC in Paris, starting in early December 2026, within a dynamic research environment at the interface between cosmology, statistics and data science.
The AstroDeep team at the APC is based in the heart of Paris and is heavily involved in international collaborations such as Euclid and the LSST DESC.
The project and the laboratory promote a collaborative, inclusive and work-life-balance-friendly working environment, with benefits such as comprehensive health cover, holiday entitlement, subsidised meals, transport discounts and optional French language courses.

• PhD, or due to defend shortly, in astrophysics, cosmology, physics, statistics or a related field.
• A good command of at least one of the following areas: cosmological simulations, large-scale structure evolution, weak gravitational shear, differentiable modelling, Bayesian inference or gradient-based inference methods.
• Experience with automatic differentiation frameworks such as JAX or PyTorch would be an advantage, as would experience with tools for cosmological simulation, Bayesian sampling or high-performance computing.
• Strong skills in Python programming and collaborative development (Git/GitHub, CI/CD, open-source practices).
• Ability to conduct research independently whilst contributing effectively to an interdisciplinary team.
• Excellent scientific communication skills in English, both written and spoken. Experience in cosmological data analysis or in large astronomical collaborations, such as DES or the LSST DESC, is an asset but not essential.