Master's Thesis: Leveraging LLM Explainers for Multi-Agent Systems

Join our Team About this opportunity: Large Language Models (LLMs) are a major driving force in artificial intelligence. Their ability to perform complex tasks has attracted significant attention and investment, as well as serious concerns among researchers and practitioners. When a model hallucinates, it can be difficult to understand why a particular answer was generated. Several explainability techniques have been developed to investigate what happens inside an LLM. Jacobian Lens (J-Lens) reads internal signals, while Natural Language Autoencoders (NLAs) reconstruct readable information about model activations. LLMs are also increasingly used to generate plans and take actions in agentic AI and multi-agent systems. Understanding their internal processes is important when relying on them for high-stakes tasks. This thesis will investigate how LLM explainability techniques can be used to identify important internal signals efficiently and provide insight into the behaviour of single-agent and multi-agent systems. The work corresponds to one or two students, 30 hp each. The location is Kista, and the preferred starting date is December 2026 or January 2027. What you will do:

• Conduct a literature review and reproduce state-of-the-art explainable AI methods for LLMs.
• Investigate, develop, and implement techniques for generating explanations for agentic AI systems.
• Study how an explanation method developed for a single agent can be extended to multi-agent systems.
• Analyse the results using available benchmarks.
• Document the findings in an academic thesis report and present the results to the research team.
• Potentially contribute to a research paper.

The skills you bring:

• You are a final-year Master's student in Machine Learning, Computer Science, or a related programme.
• You have strong theoretical knowledge of deep learning, especially generative AI and large language models.
• You have excellent programming skills in Python and PyTorch.
• You have experience with LLMs and agentic AI frameworks such as Hugging Face, OpenAI API, LangChain, LangGraph, or CrewAI.
• You have relevant coursework or project experience in deep learning, trustworthy machine learning, multi-agent systems, natural language processing, or generative AI.
• You are familiar with Git and can work independently, structure an open research question, and communicate your findings clearly.

Familiarity with Docker and Kubernetes is considered a plus. Experience with mechanistic interpretability, model evaluation, or AI safety is also beneficial. Keywords: explainable AI, large language models, agentic AI, multi-agent systems, mechanistic interpretability, and trustworthy AI. 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: 791202