Trendyol is a security instruction tuning dataset released by the Trendyol security team in 2025, designed to train advanced AI assistants for defensive network security.

The dataset contains 53,202 instruction tuning examples, covering more than 200 cybersecurity professional fields, including cloud native threats, AI/ML security, quantum computing risks, advanced incident response technologies, and other modern security challenges, providing high-quality corpus for the training of defensive security AI models.

The data includes:

• Cloud Security and Development Security Operations (DevSecOps): 18.5%

• Threat Intelligence and Threat Hunting: 16.2%

• Incident Response and Forensics: 14.8%

• Artificial Intelligence and Machine Learning Security: 12.3%

• Network and Protocol Security: 11.7%

• Identity and Access Management: 9.4%

• Emerging Technologies: 8.6%

• Specific platform security: 5.3%

• Compliance and Governance: 3.2%