Zum Inhalt springen

Expert Knowledge Transfer with Generative AI

Student Thesis, Bachelor or Master

One of the research topics in our division is Expert Knowledge Transfer. To capture knowledge that experienced engineers have never documented, we are prototyping a system for AI-assisted interviews. In the interview part, experts speak with an animated avatar in a web-based interface that listens and answers in real time. In the transfer part, we aim to improve how the system reasons, so that it detects what is missing and asks targeted questions. You can be part of this research.

Topic of the Thesis

Building prototypes. Researchers use coding assistants to build prototypes of increasing number and complexity, but often have to rebuild them before they can be used in practice. You research which development pipeline lets researchers build prototypes that a professional software team can take over and develop further without starting from scratch. You can apply and test this pipeline by building the interview interface or one of its parts.

Improving the reasoning. When experts explain their work, they skip steps that seem obvious to them. Language models tend to fill such gaps silently or to overlook them, instead of asking. You research which deterministic tools and methods from research help the system find these gaps and ask the right questions. You then measure how much each of them improves the transfer.

Choose your topic

  • Building prototypes. Researching how to build prototypes with coding assistants
  • Interview interface. The web-based interface and how experts interact with it
  • Real-time speech. Open-source speech recognition and voice output
  • Animated avatar. Generating and animating the interviewer on screen
  • Improving the reasoning. Researching methods that find what is missing
  • Gap and anomaly detection. Tools and methods that find missing steps
  • Interview techniques. Ways of asking that make expert reasoning explicit
  • Your own idea. Any other question within Expert Knowledge Transfer

What you bring

You study a subject such as computer science, human-computer interaction, psychology, cognitive science, audio technology or computational linguistics
Experience with Claude Code, VS Code, Vite, Vue, Next.js, Python or TypeScript is desirable

Contact

Kutlualp Tazefidan, 030 8104-3827, E-Mail:
Dr. Tobias Gleim, 030 8104-3168, E-Mail:

BAM, Division Safety of Transport Containers
Unter den Eichen 44-46
12203 Berlin

Requirements

  • Studies in subject-related experimental Natural Sciences e.g. Chemistry, Materials Science, Geosciences/Minearology etc.
  • At least basic Python skills are essential. Intermediate or Advanced Python is preferred;
  • Proficiency in English is NOT required.

 

Contact

CV and a letter of motivation to tomasz.stawski@bam.de; Either in DE or EN.

Dr. Tomasz Stawski
Division Structure Analysis
Phone: +49 330 8104 3395
Email: tomasz.stawski@bam.de

 

Die BAM ist eine wissenschaftlich-technische Bundesoberbehörde 
im Geschäftsbereich des Bundesministeriums für Wirtschaft und Energie.

Cookies