LLM-SE research project

As part of the LLM-SE (Large Language Model supported Systems Engineering) research project, we are working with our industry partners to develop an assistance system based on the MBSE philosophy and structure using large language models to partially automate the engineering process of warehouse technology and concepts from requirements analysis to virtual commissioning. This should interpret user input such as requirements and adaptations, transform them into individual, model-based solutions through the availability of company-specific product catalogs, historical project data and best practices and secure these through suitable validation and verification mechanisms.

Your tasks in the project

• Literature research on the modeling language SysML v2
• Research, evaluation and selection of suitable AI methods and models
• Optimization of the existing framework through the further development of a new microservice that can be integrated with the current service and offers scalable LLM-fine-tuning.
• Collecting data from industry project partners and creating pre-processing pipelines that convert raw data into ALPACA format for model training.
• Creating intuitive user interfaces that allow non-technical users to test, evaluate and refine different base models – for warehouse technologies and concepts – by using different prompts.
• Automatic generation of SysML models for warehouse technologies and concepts through the use of LLMs in the developed framework and the improvement of these models through an interactive behavior of the process engine
• The final result of your work: With the help of the framework, the layouts for different warehouse concepts including the associated warehouse technologies automatically suggested by LLMs for the corresponding material flow technology in the factory should be able to be generated automatically.

Requirements

• Very good programming skills in Python are mandatory
• Very good knowledge and practical experience in RESTful APIs is mandatory
• Very good knowledge of processing data formats such as CSV, JSON, XML and Markdown is mandatory
• Confident handling of containerization technologies such as Docker, Podman
• Practical experience with LLM fine-tuning frameworks, e.g. Unsloth, LlamaFactory is mandatory
• Practical experience with LLM inference optimization engines, e.g. Ollama, Llama.cpp or vLLM is mandatory. e.g. Ollama, Llama.cpp or vLLM is mandatory
• Practical experience with database technologies for SQL, NoSQL, vector and object databases is mandatory
• Basic knowledge of prompt engineering techniques, including one-shot, few-shot and chain-of-thought prompting is an advantage
• Basic knowledge of version control systems e.g. Git is an advantage
• Basic knowledge of front-end development e.g. React, Next.js is an advantage

Application instructions

• Start at the earliest possible date
• Please send applications by e-mail with a current transcript of records and CV as well as a short letter of motivation
• Please note that incomplete applications cannot be considered
• Further information on request by e-mail or in a personal interview

Kategorien:

Kontakt: