Industry Partner: 
Itestra

Problem: 
Itestra organises several internal events each year to foster employee connection across different locations, roles, and teams. However, manually planning interactions and seating arrangements to optimise networking and engagement is inefficient and lacks scalability. There is currently no system to systematically match employees based on relevant criteria or track previous connections. A tailored, constraint-aware application is needed to facilitate dynamic employee matching and enhance networking outcomes at events. 

Solution: 
The students developed a web-based application that enables dynamic employee matching for company events such as networking dinners and speed-meeting sessions. The tool allows organizers to import employee data, define matching constraints, and automatically generate optimized seating and interaction plans. It tracks previous matches, supports multiple event formats, and includes a flexible interface for administrators to adjust inputs as needed. This student project demonstrated how algorithm-driven planning can meaningfully improve engagement and social connectivity within a growing organization like Itestra.

GitHub: Event-Exchange-Platform

Industry Partner: 
Blekinge Institute of Technology

Problem: 
Code review processes often lack the necessary context for evaluating GUI-based tests, which results in inconsistent feedback and reduced test quality. Current review tools focus primarily on source code and do not effectively address the needs of testers reviewing front-end interactions. This forces testers to conduct reviews manually on their own machines and creates a fragmented workflow. A solution is needed that integrates GUI-relevant context into code reviews to improve collaboration, transparency, and test effectiveness.

Solution: 
The students implemented a GitHub-Action to enrich code reviews with visual and contextual information about GUIbased tests. Their tool automatically extracts relevant information, highlights changes based on structured checklists tailored to GUI testing needs. The project proved the value of context-aware review tools in improving test quality, collaboration, and feedback clarity in modern software development workflows.

GitHub: Code-Reviews-of-GUI-Tests

Industry Partner: 
Siemens

Problem: 
Problem: Big corporations invest billions in research and development annually, yet lack a centralised, accessible view of how their research efforts perform across domains. Metrics like patents, collaborations, and impact are scattered and not easily linked to strategic insights. Decision-makers lack the ability to pinpoint strengths, gaps, and opportunities for collaboration in specific research fields. A platform is needed that enables real-time analysis of an organisation's research position to support data-driven strategy in innovation management.

Solution: 
The students developed a web-based analytics dashboard to give researchers and managers on-demand visibility into the organisation's research landscape. The platform aggregates data on publications, collaborations, and key performance indicators, enabling users to explore strengths and gaps across domains. It supports slicing by topics, authors, and impact metrics, enabling informed decision-making. This work demonstrated the potential of centralised data platforms to inform decision-making and enhance

GitHub: Research-Position-Analysis-Platform

Industry Partner: 
Brivalo

Problem: 
AI-assisted code generation is becoming increasingly common, but its outputs often vary depending on phrasing, order of prompts, or even slight context changes. This unpredictability makes it difficult for development teams to rely on AI tools consistently in industrial or even regulated environments. Without consistently predictable code generation, teams face challenges in producing high-quality products and enforcing standards. A solution is needed to make AI-generated code more deterministic, consistent, and predictable—especially when used in professional software development environments.

Solution: 
The students built a prototype system that enhances the predictability of AI-generated code through structured prompts, version tracking, and output validation. It also enables teams to define coding conventions and ensure consistent AI behaviour across projects. This project showed how predictability-focused tooling can increase developer trust and make AI code generation a more reliable component of professional software development.

GitHub: Predictable-Secure-Code-Generation

Industry Partner: 
Netlight x Digital School Story

Problem: 
Digital School Story empowers students to create video content to deepen learning, but providing timely and personalised feedback at scale is a bottleneck. Manual feedback processes hinder the organisation’s ability to support a growing number of students and maintain educational quality. There is a need for an AI-based solution that can automatically evaluate student videos while complying with educational data privacy constraints. A user-friendly, maintainable, and privacy-first application would enable scalable feedback and support continued growth.

Solution: 
The students created a privacy-conscious, for nontechnical stakeholders, such as students, accessible, browserbased platform that uses AI models to generate personalised feedback on student-created educational videos. The application supports administrators in customising prompts and integrating feedback into the Digital School Story learning process. It ensures no persistent storage of student videos and adheres to strict data protection requirements. The project highlighted the feasibility of using AI to scale meaningful educational feedback while respecting privacy and resource constraints in a school context.

GitHub: Not available