Webseite Fraunhofer-Institut für Produktionstechnologie IPT

In order to automate the monotonous and cost-intensive analysis of microscopy images in cell biology, the Fraunhofer IPT is developing an automated machine learning system (AutoML) in the research project »AIxCell«. AutoML-systems train decision logics in the form of meta-models on meta-datasets in order to leverage the findings of already trained and evaluated deep learning models for new tasks or data sets.

Your tasks

  • Literature research on AutoML systems for image processing
  • Conceptual design of the meta-dataset to be stored in the Deep Learning library
  • Development of the meta-system for automatic configuration of DL-pipelines depending on the available dataset
  • Evaluation of different meta-models as decision logics (decision tree, random forest, gradient boosting, reinforcement learning etc.)
  • Evaluation and documentation of the results

Your profile

  • You are studying computer science, mechanical engineering or a comparable subject
  • You are familiar with the theory behind Machine Learning and Deep Learning (experience with AutoML is not a must)
  • You have first practical experience in the implementation of Machine Learning or Deep Learning models
  • You possess abstraction skills and conceptual competence
  • You have a high degree of personal initiative and team spirit
  • You have good language skills in German and/ or English

What we offer

  • Ideal conditions for practical experience alongside your studies
  • Collaboration in a dedicated team of scientific researchers and students within the exciting research project »AIxCell«
  • Flexible working hours and the possibility to work remotely
  • An excellent equipment of machines and devices

We look forward to receiving your application
Lars Leyendecker
Fraunhofer-Institut für Produktionstechnologie IPT
Steinbachstraße 17, 52074 Aachen
lars.leyendecker@ipt.fraunhofer.de

Um sich für diesen Job zu bewerben, sende deine Unterlagen per E-Mail an lars.leyendecker@ipt.fraunhofer.de