Thesis: Governance-Aware AI Orchestration in Public Sector Workflows
About this role
High Level Description This thesis project investigates multi-agent AI orchestration architectures tailored for public administration. The study focuses on integrating specialized Large Language Model (LLM) agents with deterministic compliance controls, auditable decision trails, and cross-institutional adaptability to automate complex administrative workflows without compromising legal sufficiency or transparency.
Project Description Modern administrative processes rely heavily on unstructured data, complex regulatory frameworks, and multi-step reviews. While single-agent LLM systems struggle with hallucinations, legal precision, and long-context compliance, modular multi-agent orchestration offers structured decomposition of administrative tasks.
This project involves prototyping and evaluating a multi-agent framework where specialized LLM agents (e.g., retrieval agents, policy critics, document extractors) collaborate under a central orchestrator. The system will process domain-specific public sector inputs (such as permit applications or policy documents) and enforce deterministic legal guardrails before presenting recommendations to human caseworkers.
Who are we looking for?
• Bachelor or Master of Science in Computer Science, Computer Engineering, Data Science, or related fields. • Graduation in June 2027 • Fluency in Swedish, both written and spoken • Fluency in English, both written and spoken
Purpose In this thesis, the candidate will investigate the following core areas:
- Evaluation of Orchestration Frameworks and Tooling • Compare state-of-the-art multi-agent orchestration frameworks regarding task routing efficiency, latency, and system reliability within public sector constraints. • Comparative Performance: Multi-Agent vs. Monolithic LLM Implementations • Measure performance, hallucination rates, and legal/policy compliance when handling complex, rule-heavy administrative workflows using modular multi-agent pipelines versus monolithic prompting. • Governance, Auditability, and Human-in-the-Loop Integration • Design and evaluate audit logging and explainability mechanisms that trace agent decision paths, tool usage, and human caseworker edits.
An Exciting Journey with Knightec Group Semcon and Knightec have joined forces as Knightec Group. Together, we are Northern Europe’s leading strategic partner in product and digital service development. With a unique combination of cross-functional expertise and a holistic business understanding, we help our clients realize their strategies – from idea to complete solution.