Fully Funded Phd And Postdoc In Machine Learning And Control Theory

Overview

The Research Training Group METEOR (Machine Learning and Control Theory for Complex Dynamical Systems), approved by the German Research Foundation (DFG), represents a flagship interdisciplinary initiative between Ludwig Maximilian University of Munich (LMU) and Technical University of Munich (TUM). This program offers 10 fully funded PhD positions and 2 postdoctoral positions commencing April 1, 2026, with contracts spanning four years for doctoral researchers and three years for postdocs at 100% TV-L E13 salary scale. METEOR addresses critical gaps in artificial intelligence by establishing formal bridges between machine learning’s data-driven approaches and control theory’s rigorous mathematical frameworks for dynamical systems, positioning itself at the forefront of next-generation AI research where safety, robustness, and theoretical guarantees are paramount.

METEOR’s research agenda is structured around four interconnected pillars. Theme T1 (Modeling and Quantification of Uncertainty for Robust Control) develops mathematical frameworks to handle stochastic disturbances and model inaccuracies in safety-critical applications like autonomous vehicles. Theme T2 (Representation for Dynamical Systems and Control) creates novel neural architectures for efficient system identification and state-space modeling of complex physical processes. Theme T3 (Control Theory for Machine Learning Algorithm Design) leverages stability analysis and optimization principles from control theory to enhance ML training dynamics, particularly for reinforcement learning and distributed systems. Theme T4 (Formal Analysis of Machine Learning Algorithms via Control Theory) establishes theoretical foundations for ML robustness against distribution shifts, adversarial attacks, and resource constraints using Lyapunov stability and robust control methodologies.

The program fosters deep methodological integration through mandatory cross-theme collaboration, requiring doctoral researchers to engage with both ML and control theory communities. Participants benefit from joint supervision across LMU and TUM, access to high-performance computing facilities, and structured training modules including specialized workshops on hybrid modeling techniques, international research exchanges, and industry partnerships with Munich’s automotive and robotics sectors. The cohort-based model emphasizes peer learning through regular retreats, journal clubs, and interdisciplinary project teams where PhD candidates and postdocs co-develop solutions to grand challenges in AI reliability.

Applications must be submitted via the LMU Graduate Center portal by November 13, 2025 (23:59 CET), requiring a CV, academic transcripts, research statement aligned with METEOR’s themes, and contact details for two references. Shortlisted candidates undergo technical interviews assessing mathematical maturity in either stochastic processes/dynamical systems (for control-oriented applicants) or optimization/statistical learning (for ML-focused candidates). While formal eligibility criteria aren’t detailed in the post, standard German academic requirements apply: Master’s degree with strong quantitative background for PhD positions, and completed PhD in related fields for postdocs, with demonstrated expertise in mathematical modeling, programming (Python/C++), and either control theory or machine learning.

Salary follows the German public sector TV-L E13 scale at 100% funding level, providing competitive tax-free stipends covering living expenses in Munich. Additional benefits include travel grants for conferences, childcare support, and access to Munich’s extensive research ecosystem including the Munich Center for Machine Learning and Bavarian AI network. The program particularly encourages applications from underrepresented groups in STEM, with dedicated mentorship initiatives and flexible working arrangements to support diverse career paths in academia and industry.

Research outcomes aim to transform AI development paradigms by establishing control theory as the missing foundation for trustworthy machine learning. Expected impacts include certified robustness guarantees for neural network controllers, energy-efficient ML algorithms inspired by control optimization, and new mathematical tools for analyzing learning dynamics in non-stationary environments. These advances hold direct relevance for autonomous systems, medical robotics, industrial automation, and climate modeling—domains where AI failures carry severe consequences and theoretical safety assurances are non-negotiable.

METEOR’s unique value proposition lies in its enforced cross-pollination between historically separate disciplines. Unlike conventional AI programs, it requires all participants to master dual technical vocabularies, with PhD candidates completing foundational coursework in both fields before specializing. Postdocs lead theme-specific subprojects while coordinating across the consortium, creating a pipeline for researchers who can seamlessly translate theoretical insights into deployable AI solutions. This model directly addresses industry pain points where ML systems fail in dynamic real-world settings due to ignored physical constraints.

Prospective candidates should note the program’s emphasis on mathematical rigor over framework proficiency. Successful applicants typically demonstrate strong backgrounds in linear algebra, stochastic processes, or optimization through publications or advanced coursework. The November 2025 deadline allows final-year Master’s students to apply before thesis completion. With Munich’s status as Europe’s third-largest technology hub, positions offer exceptional career mobility into both academic leadership roles and R&D positions at companies like BMW, Siemens, and NVIDIA that actively collaborate with METEOR.

Company:

Ludwig Maximilian University of Munich

Qualifications:

Language requirements:

Specific requirements:

Educational level:

Level of experience (years):

Senior (5+ years of experience)

Tagged as: Academia, Germany, Machine Learning, Neural Networks, NLP