Master Thesis: Device-Side Computing for Multimodal Perception and Semantic Compression

Join our Team

About this opportunity:

Ericsson Research Insight Exploration addresses the discoverability of research artefacts, and in turn researchers, by extracting and presenting links between research artefacts. This enables Ericsson researchers to discover, find, and make use of past and current work, leading to faster and better research, more impact, and less wasted work.

The objective of the thesis is to study, redesign, prototype, and evaluate the user experience and organizational value of a multi-stakeholder, cross-system, and cross-organizational AI- and knowledge graph-based scientific review and approval workflow.

The work focuses on how Ericsson Research Insight Exploration can support researchers, reviewers, approvers, and stakeholders in finding relevant prior work, identifying dependencies, understanding provenance, and making better-informed decisions. Important considerations include process transparency, algorithmic transparency, explainability of recommendations, trust in AI-based automation, appropriate levels of human control, and the risk of over-reliance on automatically generated suggestions.

What you will do:

Conduct a literature review of AI-assisted review and approval systems, human–AI interaction, explainable AI, decision support, knowledge graph-based discovery, trust in automation, and responsible AI.

Map the current external publication review and approval workflow, including stakeholder roles, information needs, decision points, handovers, data sources, and dependencies.

Investigate the current capabilities, limitations, and data sources of Ericsson Research Insight Exploration.

Conduct qualitative user research with researchers, reviewers, approvers, governance representatives, and tool owners.

Define design principles and requirements for an AI-supported workflow that follows corporate guidelines and supports responsible use of AI.

Develop or adapt a prototype workflow, interface, or system component using AI-generated suggestions, knowledge graph links, provenance information, and explanations.

Evaluate the prototype through user studies, case studies, usability testing, system metrics, and stakeholder feedback.

Analyse strengths, limitations, and organizational implications, including risks related to transparency, trust, accountability, data quality, and adoption.

Prepare and present the thesis with recommendations for integrating the solution into daily research workflows.

The skills you bring:

You are a Master's student in interaction design or human–computer interaction; computer science; information systems; machine learning, data science, or information retrieval; business and management; law with an information-technology specialisation; or a related field.

You have strong analytical, research, problem-solving, writing, and communication skills.

You are interested in qualitative user research, user-centred design, AI governance, knowledge management, or research information systems.

You are comfortable working with multiple stakeholder groups and translating needs into design requirements.

A background in academic or corporate research, such as wireless communication, is preferred but not required.