PhD and PostDoc positions in BRAID - Bridging Domains: AI-Driven Agnostic Reconstruction Frameworks

Physics World; Deutschland; Hybrid; EUR 50.000 - 65.000; Vollzeit

Zusammenfassung

The Research Consortium for Machine Learning offers full-time positions (Postdoc and PhD) across several locations in Germany. Candidates will work on detector-agnostic machine learning methodologies for high-dimensional sensor data and contribute to open science practices. Ideal candidates should have experience in machine learning, coding in Python/C++, and hold relevant academic qualifications. The role includes research collaborations, open-source contributions, and development of public datasets.

Leistungen

Multi-institutional collaboration

Travel support

Strong mentoring

Qualifikationen

• PhD or MSc in relevant fields required.
• Strong programming ability in Python and C++.
• Solid foundation in machine learning and numerics.

Aufgaben

• Research and prototype GNN/attention components.
• Contribute to open datasets and performance metrics.
• Collaborate across sites and present results.

Kenntnisse

Machine Learning for scientific data

Programming in Python

Programming in C++

Experience with PyTorch

Experience with JAX

Experience with GNNs

Experience with GPU/HPC workflows

Ausbildung

PhD in physics/CS/applied math

MSc in physics/CS/EE

Tools

Geant4

Jobbeschreibung

We offer multiple positions (Postdoc and PhD) across the newly founded BRAID consortium including KIT, RPTU, TU-Dortmund, and GSI/U Frankfurt.

BRAID develops detector-agnostic ML for irregular, high-dimensional sensor data across HEP, astroparticle, and hadron/nuclear physics. We focus on graph/transformer architectures, locality-aware adaptive dimensionality reduction, physics-informed inductive biases, open datasets with domain-specific metrics, and efficient, sustainable computing. The consortium emphasizes open-science/FAIR practices and cross-disciplinary co-design with computer science; methods will be validated on multiple experimental use cases.

Employment: Full-time, fixed-term (per host)

Start: as soon as available (2025/2026), this call will remain open until all positions are filled

Duration: 3 years

Locations covered by this call: Karlsruhe, Kaiserslautern, Dortmund.

GSI/U Frankfurt is a full member of the BRAID consortium, but positions hosted at GSI/U Frankfurt are not part of this advertisement. Recruitment for GSI/U Frankfurt will be announced via a separate call.

What you will do

• Research and prototype GNN/attention components and adaptive reduction for low-level reconstruction; integrate into domain pipelines focused on the physics research field of the respective site;
• Contribute to the release of public datasets and performance metrics and provide well-documented open-source code;
• Collaborate across sites; present results; follow good software and sustainable-computing practices.

Qualifications

• Postdoc: PhD in physics/CS/applied math (at start); strong record in ML for scientific data; experience with PyTorch/JAX, GNNs/attention, and GPU/HPC workflows;
• PhD students: MSc (or equivalent) in physics/CS/EE; solid programming (Python/C++), fundamentals in ML/numerics; motivation for ML-for-science and open science;
• Nice-to-have (all roles): simulation/Geant4; neutrino/Cherenkov or FAIR/PANDA/CBM experience; uncertainty quantification; open-source contributions.

We offer

a multi-institutional environment with regular consortium meetings/workshops, travel support, strong mentoring, open-science culture, and access to GPU/HPC resources.

Application instructions (common to the consortium)

• Your preference/constraints for a specific site if you have any.
• CV (includes talk and publication lists).
• Research interests (max 1 page).
• Links to code/open-source (optional). No formal cover letter.
• Additionally, 2 (PhD) or 3 (Postdoc) reference letters to be sent directly by the referees (see contacts).
• Equal opportunity: We value inclusivity and strongly encourage applications from under-represented groups. We adhere to FAIR/open-science and sustainable-computing principles.

Applications and reference letters to be sent to:

Ulrike Hahn ([e-mail])

Contacts for questions about the consortium (PIs):

Jan Kieseler (KIT, [e-mail])

Nicolas R. Gauger (RPTU, [e-mail])

Christian Glaser (TU Dortmund, [e-mail])

Anastasios Belias (GSI/U Frankfurt, [e-mail])