Summary from listing
The Master's thesis will design and implement alternative multi-agent architectures for AI-assisted engineering report generation and experimentally evaluate their effectiveness. The student will extend the HOMER research platform, explore agent decomposition, coordination, retrieval, reasoning, synthesis, verification, and orchestration strategies, and assess architectures against criteria such as factual correctness, completeness, consistency, traceability, robustness, and computational cost. The work will use large language models, RAG, embeddings, vector databases, Python, and frameworks such as LangChain or LangGraph, with research conducted in collaboration with SCK CEN using technical documentation from the MYRRHA and LEANDREA programmes.
