PROVEX: A pressure-based learning system for evidence-driven mastery in computer science education

The proliferation of passive learning tools, AI-generated solutions, and hint-driven coding platforms has created a critical gap between perceived and actual technical competence among learners. This paper presents PROVEX, a pressure-based adaptive learning system designed for students in computer science and information technology programs. PROVEX enforces mastery only when learners demonstrate consistent, clean, independent problem-solving ability under controlled constraints. The system restricts passive assistance, introduces structured failure as a learning mechanism, and employs LLMs exclusively for content generation and analysis, not for tutoring or hint delivery. This paper describes the system architecture, pedagogical foundations, requirement analysis, and proposed solution framework. The design is grounded in established research on constraint-based learning, desirable difficulties, and evidence-based assessment. PROVEX addresses three core requirements: preventing superficial progress, supporting absolute beginners without fostering dependency, and leveraging LLMs without creating over-reliance. The system is evaluated against existing platforms including LeetCode, HackerRank, and Codecademy, and demonstrates conceptual superiority in enforcing genuine competency validation.