Applied AI Researcher
- Organization
- Fraunhofer IAO
- Location
- Stuttgart, Germany
- Duration
- 1+ yrsCurrent
Highlights
- Technical lead of the research line LLM-Based Human Digital Twins for Research and Decision Making: SozioMimic, MindsetTwin, and the Twinlyses prototype
- AI demonstrators for AI-Summit and Fraunhofer industry events
- AI strategy consultation to SMEs with the Baden-Württemberg Ministry
Applied AI research role focused on developing LLM and multi-agent solutions, co-creating industry prototypes, consulting SMEs, and translating emerging AI methods into practical demonstrators for business and public-sector stakeholders.
Work and responsibilities
- Developing and implementing applied AI solutions based on large language models, multi-agent systems, retrieval workflows, and context-aware agent architectures.
- Designing AI demonstrators for AI-Summit and Fraunhofer events, turning research concepts into working prototypes that communicate value to industry stakeholders.
- Partnering with companies across mobility, market research, UX, and enterprise domains to co-create practical AI solutions from ambiguous problem statements.
- Providing AI consultation to industries and SMEs, including collaboration with the Baden-Wuerttemberg Ministry on tailored solutions that help local businesses adopt and leverage AI effectively.
- Managing small to medium-sized applied research projects, coordinating contributors, aligning technical direction, and balancing prototype delivery with research quality.
- Conducting knowledge-sharing sessions for Technology Management master's students and industry professionals in Baden-Wuerttemberg to raise awareness of AI capabilities, risks, and adoption paths.
- UrbanClimateAI: Developing AI architecture, backend/frontend components, and an initial MVP for source-grounded climate and urban intelligence at asset level.
- Office Simulation: Developing the Mesa-based agent environment, simulation architecture, backend, and MVP interface for comparing workplace scenarios with repeated Monte Carlo runs.
- LLM-Based Human Digital Twins for Research and Decision Making: Leading the research line that connects SozioMimic and MindsetTwin, and building its prototype, Twinlyses: consent-governed human digital twins grounded in contributors' own evidence, with a contributor app and a research workspace.
- MindsetTwin: Extended the SozioMimic research line toward behavioral digital twins that model cognition, preference formation, and subjective decision-making for advanced market research (Nov 2025 – Mar 2026).
- Conversational Analytics Platform: Applied these responsibilities to conversational intent analysis, behavioral insight extraction, and explainable UX recommendations for mobility-domain assistant interactions.
- SozioMimic: Implemented LLM-driven generative agents that simulate human attitudes, preferences, and social context from qualitative data for market research.
Methods and tools
Languages & App Stack
- Python
- TypeScript
- SQL
- SvelteKit
Agentic AI
- Agentic workflows
- Context engineering
- Memory management
- LangChain
- LangGraph
- CrewAI
Data & Retrieval
- LanceDB
- Neo4j
- Graph RAG
- Semantic retrieval
- Embedding workflows
Methods
- Fine-tuning
- Generative personas
- Conversational analytics
- Behavioral simulation
Domain Work
- Market research
- Mobility UX
- Qualitative data analysis
- Cognitive science