Computer Science Correspondence Seminar (Korespondenční seminář z informatiky) is a competition organised by FI MU students aimed at secondary school students. It prepares the participants for studies at the faculty, and successful candidates can be granted enrollment without admission. The evaluation system for the seminar, however, has evolved little over time and exhibits limitations in security, extensibility and maintainability.

The goals for this thesis are:

• Analyse the current evaluation system and identify its limitations with respect to scalability, usability and security.

• Review existing approaches to automated evaluation of programming assignments in educational settings.

• Design a prototype framework to support automated evaluation for the seminar's needs. Propose and describe improvements to security and maintainability.

• Demonstrate the prototype's functionality on real submissions and its extensibility by adapting it for a different setting, e.g. a programming seminar taught at the faculty. Discuss the improvements and limitations of the proposed solution compared to the existing solution.