PhD Position in Scientific Machine Learning & Data Science at University of Munich, Germany

Summary

The Technical University of Munich (TUM) is offering a fully funded doctoral position (100% TVL E13) within the HDSC/MDSI Tandem Project titled “Data science at scale: Training neural models with mixed precision solvers.” This interdisciplinary role is hosted jointly by Prof. Dr. Michael Bader, Prof. Dr. Felix Dietrich, and Prof. Dr. Hartwig Anzt at TUM CIT. The selected researcher will focus on developing scalable linear solvers and High Performance Computing (HPC) algorithms to train neural models on large datasets for solving Partial Differential Equations (PDEs).

PhD Position in Scientific Machine Learning & Data Science at University of Munich, Germany

Designation

• Position: Doctoral Researcher / Academic Staff (PhD Candidate)
• Degree Targeted: Dr. rer. nat. (“German PhD”)
• Pay Scale: 100% TVL E13 (German public sector rate)

Overview

Research Area

• Scientific Machine Learning (SciML)
• Physics-Enhanced Machine Learning
• High-Performance Computing (HPC) & GPU Architectures
• Linear Algebra & Mixed Precision Solvers
• Computational Mathematics & Partial Differential Equations (PDEs)

Location

• Primary Workplace: TUM Campus Garching, Munich, Bavaria, Germany.
• Collaboration: Secondary interaction and semesterly research visits with researchers at the Heilbronn campus.

Eligibility / Qualification

Required Qualifications

• Degree: Master’s degree in Informatics, Mathematics, or a related field (e.g., Computational Science/Engineering).
• Domain Knowledge: Strong background in machine learning, Scientific Computing (particularly linear algebra and numerics), and High-Performance Computing (HPC).
• Technical Skills: High proficiency in programming with experience in GPU architectures, C++, and Python.
• Language & Soft Skills: Excellent English communication and collaboration skills, analytical thinking, structured work habits, and high intrinsic motivation.

Preferred Qualifications

• Prior experience with linear operators, eigenproblems, or the design and implementation of numerical solvers.

Job Description

• Develop general-purpose, scalable linear solvers for training neural networks that represent solutions to partial differential equations (PDEs).
• Build scalable HPC algorithms and mixed-precision solvers capable of training neural models on massive datasets exceeding single-node capabilities.
• Formulate efficient iterative algorithms for training networks based on random features applied to regression, classification, and time-dependent PDEs (such as large-scale tsunami simulations).
• Utilize GPU code generators for matrix multiplication kernels and integrate software for seamless use in Python.
• Collaborate closely with interdisciplinary researchers across Garching and Heilbronn campuses, participating in research meetings and visits bridging computational mathematics and physics-enhanced machine learning.

How to Apply

1. Fill out the application via the official TUM SCML Registration Form.
2. In the free text field of the form, explicitly state: Application: MDSI/HDSC 2026.
3. Upload one single PDF document containing the following materials:
   • Motivation Letter (up to 1 page): Explaining why you want to work on this project, pursue a PhD in this field, and join this team at TUM.
   • Curriculum Vitae (CV): Detailing academic education, optional industry experience, and optional completed projects.
   • Transcript of Records: Including all completed courses and grades.
   • Reference Contact: Contact details of one reference person (will only be contacted if the hiring decision requires further clarification).

Last Date to Apply

• Deadline: Friday, August 14, 2026 (EOD)

Link