Thesis: Governance-Aware AI Orchestration in Public Sector Workflows

Public administration involves unstructured information, detailed regulations, and reviews across several stages, which can challenge single-agent LLM systems. This thesis explores how coordinated, specialized AI agents can support administrative workflows while maintaining legal compliance and transparency. The focus is on architectures that combine deterministic compliance checks, traceable decisions, and adaptability across public institutions.

The assignment

• Build and evaluate a prototype in which a central orchestrator coordinates specialized LLM agents for tasks such as information retrieval, policy review, and document extraction.
• Process public-sector material, such as permit applications or policy documents, and apply deterministic legal checks before recommendations reach human caseworkers.
• Compare multi-agent orchestration frameworks in terms of task routing efficiency, latency, and reliability under public-sector constraints.
• Assess modular multi-agent pipelines against monolithic LLM prompting, measuring performance, hallucination rates, and adherence to laws and policies.
• Develop and assess audit logs and explainability mechanisms that capture agent decisions, tool use, and changes made by caseworkers.

Who we are looking for

• Studying towards a bachelor's or master's degree in Computer Science, Computer Engineering, Data Science, or a related discipline.
• Expected to graduate in June 2027.
• Fluent in both spoken and written Swedish and English.

What we offer

• An opportunity to undertake a thesis at Knightec Group, which brings together Semcon and Knightec and combines cross-functional expertise in product and digital service development.

Practical details

Based at the Sundsvall office. Start in January or March 2027. Apply through the company's careers page as soon as possible, no later than 30 November 2026. Questions can be sent to johanna.edstrom@knightec.se.