Cybersecurity LLM Fine-tuning Data Annotator
Fine-tuningI contributed to the fine-tuning of a Cybersecurity Large Language Model (LLM) for code and vulnerability analysis tasks. The project involved supervised domain-specific instruction and reinforcement to optimize the LLM's security, accuracy, and automation capabilities. I focused on preparing curated code datasets and annotating them for vulnerability detection, automated penetration testing, and multi-agent task orchestration. • Implemented prompt engineering and SFT (Supervised Fine-Tuning) methods for code examples. • Designed instructions and responses for code analysis and security testing. • Evaluated model output accuracy and provided RLHF-based feedback. • Utilized domain knowledge in cybersecurity to craft annotated data and prompts.