Call for applications for the selection process to fill a junior postdoctoral researcher position in machine learning applied to astrophysics

The purpose of this call is to select a candidate to fill a junior postdoctoral researcher position assigned to the Research Area of the Institute of Space Studies of Catalonia (IEEC), within the framework of the SPOTLESS project.

SPOTLESS is a project funded through an ERC Advanced Grant that aims to develop a comprehensive strategy to model and mitigate the impact of stellar activity on the signals used to detect Earth-like exoplanets and characterise the atmospheres of small planets.

The project combines physical models of stellar surfaces, spectral simulations, high-precision observations and machine-learning techniques. The data used include photometry, radial velocities, high-resolution spectra and transmission spectroscopy obtained with instruments and missions such as CARMENES, HARPS, ESPRESSO, JWST, PLATO and Ariel.

One of the central research lines of SPOTLESS is the development of machine-learning methodologies trained on physically realistic simulations produced with StarSim. These tools are intended to identify, predict and correct signals caused by stellar magnetic activity, separating them from planetary signals. The recruitment of a researcher with experience in machine learning applied to astrophysics is required to expand, validate and apply these methodologies to new datasets and scientific cases.

• Positions to be filled: 1. Call for applications for the selection process to fill a junior postdoctoral researcher position in machine learning applied to astrophysics
• Professional group: Group III. Scientific staff
• Professional category: Junior postdoctoral researcher
• Working hours: 37,5 h/week
• Workplace location: Institute of Space Sciences, ICE-CSIC, UAB Campus, 08193 Bellaterra

The selected candidate will join the SPOTLESS project team and contribute to the development of machine-learning methodologies for the analysis and correction of stellar variability in exoplanet observations.

The work will focus on the design, training, validation and interpretation of machine-learning models applied to heterogeneous astrophysical data, including photometric and spectroscopic time series, radial velocities, spectral line profiles and transmission spectra. The models will be trained using a combination of real data and simulations generated with StarSim.

The selected candidate will have the following duties and responsibilities:

• Develop, train and validate machine-learning models applied to stellar variability and exoplanet detection and characterisation.
• Explore neural-network architectures suitable for astrophysical time series, including convolutional and recurrent networks, attention mechanisms and transformers.
• Develop methodologies to integrate heterogeneous, multiband, multi-technique, irregularly sampled or non-simultaneous data.
• Use simulations generated with StarSim to build physically realistic training datasets representative of different stellar types and activity levels.
• Develop techniques to separate stellar signals from planetary signals in radial-velocity, photometric and transmission-spectroscopy data.
• Contribute to the improvement and extension of methodologies developed within the project, such as CANSTAR and other stellar-activity correction tools.
• Study the robustness, interpretability and generalisation capability of the models, and quantify the uncertainties of their predictions.
• Identify the photometric and spectroscopic observables with the greatest predictive power for characterising stellar activity states.
• Apply the developed methodologies to data from CARMENES, HARPS, ESPRESSO, JWST, PLATO and other relevant facilities.
• Collaborate with team members specialising in physical modelling, solar physics, stellar atmospheres and exoplanet analysis.
• Participate in the development, documentation, verification and maintenance of the project’s scientific software.
• Prepare scientific articles for international peer-reviewed journals and present the results at scientific conferences and meetings.

The following criteria will be assessed during the selection process:

• ­Education: Applicants must hold a PhD in Physics, Astrophysics, Astronomy or a closely related scientific discipline.

The following aspects will be considered desirable merits (maximum 20 points):

• ­Experience in the development and application of machine-learning methodologies to astrophysical problems: up to 6 points.
• Experience with neural networks and architectures for time-series analysis, such as convolutional networks, recurrent networks, transformers or attention mechanisms: up to 4 points.
• Experience in the analysis of astrophysical data, particularly photometry, spectroscopy, radial velocities or transmission spectroscopy: up to 3 points.
• Knowledge of statistics, inference, uncertainty quantification, interpretability or validation of machine-learning models: up to 3 points.
• Scientific software development skills, particularly in Python and machine-learning libraries such as PyTorch, TensorFlow, JAX or equivalent frameworks: up to 2 points.
• Track record of scientific publications and experience in international collaborations: up to 2 points.

Interested persons who meet the established requirements can send two reference letters, cover letter and the personal curriculum in PDF format, signed as certification, with a brief description of the tasks carried out in the positions previously occupied, to the email address: RecursosHumans@ieec.cat indicating the reference "Surnames and Name – OF_IEEC_262_2026" in the subject line of the email.