The successful candidate will be employed under a regular employment contract.
The position is to be filled at the earliest possible date and offered for a fixed term for two years initially.
A one-year extension is planned. A further extension is desired.
The fixed-term employment is possible as it constitutes one of the fixed-term options of the Wissenschaftszeitvertragsgesetz (German Act on Fixed-term Scientific Contracts).
This is a full-time position.
The successful candidate has the opportunity to pursue a doctoral degree in this position.
The salary is based on the German public service salary scale (TV-L).
The position corresponds to a pay grade of EG 13 TV-L.
Artificial intelligence is fundamentally changing the way people work with complex software. At MMI, we develop multimodal AI assistants that guide domain experts through the complicated processes and tasks in the context of systems engineering, modelling and simulation in various application areas – from the initial idea through system structure, modelling and simulation to evaluation. Instead of having to operate a complex simulation interface, users describe their request in natural language; the assistant provides the required simulation expertise. In this way, simulation-based methods become accessible even to companies that have so far lacked the expertise or capacity.
At its core lies the combination of multimodal large language models and vision-language models with tool-specific knowledge via Retrieval-Augmented Generation, hybrid semantic search, embeddings, and vector databases. This is complemented by secure, tool-based access to real applications through the Model Context Protocol with structured, schema-valid outputs (constrained decoding), validation against metamodels and ontologies, and guardrails to safeguard system states. The overall framework is provided by agentic architectures that decompose complex requests through reasoning and planning into hierarchical subtasks, orchestrate several specialised agents, and resolve them in the interplay of human and agent (human- and agent-in-the-loop).
We are looking for a research associate (f/m/d) with the intention to pursue a PhD and a strong interest in applied artificial intelligence: in LLM-based assistance and agent systems, in knowledge provisioning, and in connecting language models with real tools and 3D simulation models. The focus is on developing practically applicable systems that support users productively.
We offer:
- The opportunity to pursue a PhD in a highly topical research field at the intersection of generative AI, agent systems, systems engineering, and 3D simulation
- A full-time position (remuneration according to TV-L E13)
- Collaboration in an externally funded research project with industry partners and a concrete application context
- The development of sophisticated agentic AI systems (LLMs, RAG, tool use/MCP, multi-agent orchestration), not merely prototypical demonstrators
- Access to state-of-the-art research infrastructure, in-house simulation tools, and Digital Twins
- A creative, interdisciplinary research environment at RWTH Aachen University, one of Europe’s leading technical universities and a University of Excellence
- Scope to develop your own ideas and research directions
Up to 40% remote work within the framework of an existing works agreement
-
About us:
The Institute for Man-Machine Interaction (MMI) is a nationally and internationally active and successful research institution within the Faculty of Electrical Engineering and Information Technology of RWTH Aachen University. MMI specialises in the modelling, simulation, control, and analysis of complex systems in their operational environment, in the interaction and networking of machines, devices, and people, and in intelligent systems for space applications and on Earth. Digital Twins realised at MMI revolutionise development processes and the operation of (not only) robotic systems. They form the basis for the convergence of the virtual and the real world – across industries and disciplines. The diverse applications range from planetary rovers and modular satellite systems to the “International Space Station”, from industrial robots to “Virtual Production”, and from wheel loaders to the “Virtual Forest”.
If you have any questions about the position or the project, we look forward to hearing from you – informal preliminary enquiries are expressly welcome.
What you bring:
- An excellent Master’s degree (or comparable) in Computer Science, Data Science, Artificial Intelligence, Electrical Engineering, Computer Engineering, or a related field
- Very good programming skills in Python as well as experience with common ML/LLM frameworks
- A sound understanding of modern AI methods, in particular large language models (LLMs)
- Interest in applied AI: RAG, tool calling, agent architectures, and prompt and context engineering
- An independent, structured, and research-oriented working style
Very good German and good English language skills
-
Desirable:
- Experience with agent frameworks and their orchestration (e.g., multi-agent systems, ReAct, planner-executor), the Model Context Protocol (MCP), or tool/function calling
- Experience with Retrieval-Augmented Generation, hybrid semantic search, re-ranking, embeddings, and vector databases
- Experience in adapting and operating language models
- Knowledge of evaluating LLM systems (benchmarking, automated assessment) as well as LLMOps
- Knowledge of 3D simulation, Digital Twins, robotics, or automation engineering, as well as knowledge-based methods (ontologies, knowledge graphs, metamodels)
- Experience with containerisation, version control, and CI/CD (e.g., Git, GitLab, Docker)
- Interest in operating language models on premises (local LLMs, data protection)
- Research and development of a multimodal AI assistant that translates natural-language requests into secure actions on real 3D simulation models
- Design and implementation of agentic architectures with reasoning, planning, and hierarchical task decomposition, orchestration of multiple agents, and human- and agent-in-the-loop
- Integration of large language models and of tool-specific knowledge via Retrieval-Augmented Generation (RAG), embeddings, and vector databases
- Development of tool-agnostic interfaces to applications via the Model Context Protocol (MCP) with structured, schema-valid outputs and safeguarding of system states against metamodels
- Design of methods for reliability and traceability (guardrails, error detection and correction, explainability)
- Establishment of a systematic evaluation based on technical target metrics (e.g., task success, validity of results, efficiency gains)
- Collaboration in research and industry projects with a strong practical focus
- Supervision of student theses and contribution to teaching
RWTH is proud to be certified as a family-friendly university. As part of our Campus Health initiative, RWTH offers a wide range of health, counseling, and prevention services, including university sports. Employees under collective bargaining agreements and civil servants also benefit from a comprehensive continuing education program and the option to purchase a job ticket for convenient commuting.
RWTH is an equal opportunities employer. We welcome and encourage applications from all qualified candidates and make employment decisions without regard to national or ethnic origin, sex, sexual orientation, gender identity, religion, disability or age. RWTH is especially committed to supporting the advancement of women in academia. When women are underrepresented in a university unit, preferential consideration will be given to female applicants among candidates of equal merit and professional standing, unless compelling reasons favor another individual.
In keeping with RWTH’s commitment to equality of opportunity, we ask that you do not include a photo with your application.
Please refer to https://www.rwth-aachen.de/applicant-privacy-policy for information about how we collect and process applicant data in accordance with Articles 13 and 14 of the European Union’s General Data Protection Regulation (GDPR).
EG 13 TV-L
