Master's Thesis: Verification of Reasoning Traces for LLM-Based Agents

Join our Team About this opportunity: LLM-based agentic workflows can execute highly complex tasks with strong performance, but their intermediate reasoning processes and outcomes—derived from adaptive behaviour and external tool interactions—can be challenging to verify and validate. Existing techniques for traditional AI and machine-learning models face limitations when providing sound and complete verification of decisions and outcomes produced by these emerging agents. The complexity increases in real-world scenarios involving orchestrated, system-level workflows. This thesis will investigate structured and formally verifiable methods for evaluating the acceptability of reasoning chains through supporting, contradictory, and unresolved claims, while establishing the contextual relevance of the reasoning process. The objective is to develop a hybrid approach for verifying context-supported, step-level reasoning traces in tool-using agents. The work corresponds to one or two students, 30 hp each. The location is Stockholm/Kista, and the preferred starting date is early January. What you will do:

• Conduct a focused literature review of techniques for verifying multi-step reasoning traces in agentic workflows with tool interaction.
• Design and develop a methodology for obtaining and analysing step-level reasoning traces.
• Derive measures for semantic grounding, contextual relevance, and coherence of the obtained traces.
• Develop and evaluate a hybrid verification approach using suitable symbolic and statistical methods.
• Analyse results and document limitations.
• Formulate recommendations for future formalisation and trustworthy deployment of LLM-based agents.

The skills you bring:

• You are studying Computer Science, Data Science, Computer Engineering, Electrical Engineering, or a related field.
• You have solid Python programming skills.
• You are interested in careful experimental design, formal reasoning, and evaluation.
• You have good analytical, problem-solving, and technical writing skills.

The following knowledge or experience is considered a plus:

• Machine reasoning, including argumentation, formal logic, or constraint optimisation.
• Machine learning techniques and large language models.
• Agentic AI frameworks and tool-using agents.
• Trustworthy AI and responsible AI.
• Statistics, formal verification, autonomous networks, or related areas.

Keywords: machine reasoning, formal verification, constraint optimisation, statistics, Python, AI agents, and autonomous networks. Why join Ericsson?At Ericsson, you´ll have an outstanding opportunity. The chance to use your skills and imagination to push the boundaries of what´s possible. To build solutions never seen before to some of the world’s toughest problems. You´ll be challenged, but you won’t be alone. You´ll be joining a team of diverse innovators, all driven to go beyond the status quo to craft what comes next. What happens once you apply?Click Here to find all you need to know about what our typical hiring process looks like.Encouraging a diverse and inclusive organization is core to our values at Ericsson, that's why we champion it in everything we do. We truly believe that by collaborating with people with different experiences we drive innovation, which is essential for our future growth. We encourage people from all backgrounds to apply and realize their full potential as part of our Ericsson team. Ericsson is proud to be an Equal Opportunity Employer. learn more. Primary country and city: Sweden (SE) || Stockholm Req ID: 791199