Master Thesis: Requirements to Verified RTL Releases: A Multi-Agent Traceability Framework for ASIC

Complex ASIC IP development requires reliable links between requirements, engineering artifacts and release evidence across agile, cross-functional teams. This thesis at Ericsson will develop and evaluate a multi-agent framework for feature-level traceability and quality assurance of internal and external releases. The aim is to strengthen team autonomy while maintaining system consistency, transparent releases and engineering accountability.

The assignment

• Build a prototype that links requirement specifications to Jira work-item progress, RTL changes, documentation, reference models, verification evidence, FPGA builds and release baselines.
• Define specialized agent responsibilities and implement a coordinating agent to create an auditable artifact graph, assess change impacts and produce explainable quality reports.
• Combine LLM-based semantic reasoning with deterministic rules for versions, dependencies and release quality gates.
• Evaluate the prototype through traceability accuracy, warning precision, evidence completeness, unsupported-claim rate and practitioner feedback.

Who we are looking for

• Knowledge of computer science, electrical engineering or a related discipline.
• Experience with at least one programming language; Python is not required.
• Curiosity and motivation to learn unfamiliar concepts.
• Machine learning experience is an advantage.

Practical details

Master’s thesis at Ericsson in Lund.