Certiguard: a fine-tuned LLM for mastering security certification exams

Cybersecurity certification exams have become an important pathway for developing professional competence, yet preparation for exams such as CISA, CRISC, CEH, and CCNA often relies on fragmented materials, costly training programs, and limited individualized feedback. This study introduces CertiGuard, a resource-efficient cybersecurity training framework built on LLaMA-2-7B. The model is adapted through a two-stage process: continued pre-training on a curated cybersecurity corpus, followed by supervised fine-tuning with certification-aligned instructions. To reduce the computational cost of domain adaptation, the framework applies QLoRA-based fine-tuning and organizes model use through beginner, intermediate, and advanced proficiency tiers. Experiments were conducted on four certification-oriented question-answering datasets covering auditing, risk and control, ethical hacking, and networking domains. The base Certibase model achieved an average accuracy of 54.1%, while the best structured prompt improved the average accuracy to 72.0%. Further evaluation of reasoning strategies showed that self-consistency Chain-of-Thought achieved the strongest overall performance, reaching an average accuracy of 86.5% across the four exam domains. The results suggest that domain adaptation, structured prompting, and reasoning-oriented inference can substantially improve LLM performance on cybersecurity certification questions. At the same time, the study highlights practical limitations related to small evaluation sets, dataset provenance, copyright restrictions, and the need for broader human-centered evaluation. Overall, CertiGuard provides a lightweight and scalable approach for certification-oriented cybersecurity training, while offering a basis for future adaptive learning systems and reproducible evaluation benchmarks.