CV Dr. Peter Gmeiner

Senior Machine Learning Engineer @GfK, Mathematician @AlgoBalance
Github, LinkedIn, Xing, Google Scholar, Twitter: @PeterGmeiner4,
E-Mail: peter.gmeiner(at)
last update: 05/2022


A mathematician who is currently working in ML/AI, applying causal inference in practice (e.g. implementing a causal inference engine). I have years of experience in developing software as a full-stack software developer (Python, R, Java, C++, C#, SQL, Jupyter, PHP, HTML, Docker, Kubernetes, Spark, UML, ...). This technical knowledge is built on top of a foundational understanding of Machine Learning and AI techniques supplemented with years of research experience at university. My research interest is in causal inference, causal discovery, and in the connection between ML and causality.


since 02/2019    Senior Machine Learning Engineer at GfK SE, Nürnberg, Germany
  • Development, implementation, and application of causal inference and causal discovery algorithms.
  • Proof of concept for root cause analysis of observed market changes.
  • Design and development of prototypes and proof of concepts in the Data Science and Machine Learning environment.
  • Development of market simulation and optimization models.
  • Development of a rule-based recommendation algorithm on basis of market specific key figures.
  • Scaling and productionizing of data science and machine learning applications as microservices.
Used technologies: Python, R, Jupyter, PySpark, Java, Scrum, Git, Docker, Hive, SQL, NoSQL DB's, microservice architecture.
11/2016-12/2018    Software engineer at Orpheus GmbH, Nürnberg, Germany
  • Development and application of (machine learning) algorithms for automatic document classification and for searching similar companies.
  • Development of crawling methods for data enrichment.
  • Development of client-server applications.
  • Co-development of a cloud infrastructure.
  • Application of Deep Learning for text classification.
Used technologies: Java, JUnit, Spring, Hibernate, HTML, Python, Scrum, Git, Docker, Helm, Kubernetes, Keycloak, Vaadin, Keras, Apache Spark.
since 04/2016    Managing director, software, and algorithm developer at AlgoBalance UG (haftungsbeschränkt), Germany
  • Requirements analysis, data model, design, and concept of software solutions.
  • Algorithm development and implementation.
  • Implementation of web applications and mobile applications.
  • Organizational and administrative tasks within the management.
Used technologies: Python, HTML, PHP, Java, MySQL, Javascript, jQuery, C#, R, UML, ER diagrams, Apache, Linux, Android, iOS, Git, Docker.
04/2016    Co-founder of AlgoBalance UG (haftungsbeschränkt), Germany
10/2015-09/2016    EXIST scholarship holder, idea provider and know-how provider for the 'FoodOptimizer' project.
08/2015-09/2015    Analysis algorithmic specialist at codemanufaktur GmbH, Erlangen, Germany
  • Development of a knowledge-based recommendation algorithm.
  • Prototyping.
Used technologies: Java, MySQL, ER diagrams.
11/2010-03/2015    Research assistant at the Department Mathematics of the University of Erlangen-Nürnberg, Germany
  • Designing student exercises and organizing courses.
  • Teaching large exercise classes up to 350-450 students.
  • Supervision of tutorial groups and tutors.
  • Planning of pre-course, exam, post-exam and vacation courses.
  • Talks at the Mathematical Research Institute Oberwolfach, the ECCS 2012 in Brussels, the University of Chemnitz, and the Max-Planck Institute Leipzig.
Used technologies: Latex, Matlab, Mathematica, Maple, R.


   PhD in Mathematics, University of Erlangen-Nürnberg,
Thesis: 'Spectral Hypergraph Partitioning and Relative Entropy'
   Diploma in Mathematics, University of Erlangen-Nürnberg,
Thesis: 'Komplexitätsmaße und Emergenz'
   intermediate diploma in computer science, University of Applied Sciences Nürnberg
   subject-related advanced technical college entrance qualification (Fachabitur)


   government grant (EXIST) for the software project 'FoodOptimizer' (now called Nutrimizer)


   P. Gmeiner: Information-Theoretic Approximation to Causal Models, arXiv:2007.15047, github, 2020.
   P. Gmeiner: Lower and Upper Bounds for the Kullback-Leibler Divergence between Hierarchical Models, Preprint, 2014.
   P. Gmeiner: Some Properties of Persistent Mutual Information, In: Proceedings of the European Conference on Complex Systems 2012, Springer Proceedings in Complexity 2013, pp 867-876, arXiv.
   P. Gmeiner: Vertex and edge coloring models, Oberwolfach Reports (OWR) vol. 16, 2013.
   P. Gmeiner, M. Seri: The classical KAM Theory, Oberwolfach Reports (OWR) vol. 9, 2012.
   P. Gmeiner: Quantum ergodicity for quantum graphs, Oberwolfach Reports (OWR) vol. 49, 2011.
   P. Gmeiner, A. Mohapatra: On the dynamics of a discrete Brusselator model, Project report at the Internationalen summer school Göttingen with topic: 'dynamical systems', 2011.
   P. Gmeiner: Equality conditions for internal entropies of certain classical and quantum models, arXiv:1108.5303, 2011.
   P. Gmeiner: Spectral Hypergraph Partitioning and Relative Entropy, Dissertation, 2015.
   P. Gmeiner: Komplexitätsmaße und Emergenz, Diplomarbeit, in german, 2010.


   On (un)observed confounders in causal models, Talk at GfK Nürnberg.
   On (un)observed confounders in causal models, Causal Inference Meeting.
   Causal Effect Estimation with Double Machine Learning, Causal Inference Meeting.
   Causal Effect Estimation with Double Machine Learning, we.create 2022, GfK Nürnberg.
   Causal Inference Engine, Talk at GfK Nürnberg.
   Statistical Transportability between Markets, we.create 2021, GfK Nürnberg.
   Causal discovery with Point of Sales data, 3rd International Conference on Advanced Research Methods and Analytics (CARMA 2020), 2020.
   Causal discovery with POS data and business knowledge, we.create 2020, GfK Nürnberg.
   Causality: Applications and Connections to Machine Learning, Talk at GfK Nürnberg.
   Information-Theoretic Cheeger Inequalities, Special Seminar, MPI Leipzig.
   Vertex and edge coloring models, 'Arbeitsgemeinschaft' with topic 'Limits of Structures', Oberwolfach.
   Some Properties of Persistent Mutual Information, European Conference on Complex Systems 2012 in Brussels.
   Spektren von Laplaceoperatoren auf Hypergraphen, Forschungsseminar Analysis, Stochastik und Mathematische Physik, TU Chemnitz.
   The classical KAM Theory, 'Arbeitsgemeinschaft' with topic 'Quasiperiodic Schrödinger Operators', Oberwolfach.
   Überraschende Geometrie und Fraktale, Tag der Mathematik, Erlangen.
   Quantum ergodicity for quantum graphs, 'Arbeitsgemeinschaft' with topic 'Quantum Ergodicity', Oberwolfach.


Mathematical skills
    graph theory, information theory, statistical inference, causal inference, statistics and probability theory, exponential families, Bayesian networks, calculus, algorithm development and analysis, optimization
Methodological skills
    (log)-linear models, regression models, random forest, SVM, neural networks (especially deep learning), LDA, reinforcement learning, clustering methods, causal models, causal inference and discovery methods, GAM, hidden Markov models, natural language processing (Word2Vec, Fasttext), (non)-linear optimization, particle swarm optimization
Programming languages
    C, C++, C#, SQL, Java, Visual Basic, HTML, PHP, Python, Javascript, jQuery, Latex, Matlab, Mathematica, Maple, R
Application software
    Office (MS Office, OpenOffice), Git, Dia, Apache
Development environments
    Eclipse, Visual Studio, Aptana Studio, RStudio, Xamarin Studio, Intellij IDEA, PyCharm, Jupyter, Hive
Software design
    UML, design patterns, object orientation, microservice architecture
Operating systems
    MS Windows, Linux (Debian), MacOS, Android, iOS
Project management
    Scrum, Agile Software Development
Language skills
    German (native), English (fluent)
Other interests and hobbies
    running (active), hiking, artificial intelligence, nutritional sciences, philosophy