安全工程师（大模型安全能力提升方向） - 火山方舟
Core
Enhancing large language models' capabilities in cybersecurity tasks including vulnerability discovery, code auditing, penetration testing, and reverse analysis.
Role type
Senior IC security engineer specializing in LLM safety
Builds
Production ML models for cybersecurity
Domain
Cybersecurity + Large Language Models
Deliverable
production ML models
Required skills
vulnerability discovery, code auditing, penetration testing, reverse engineering, exploit development, attack surface analysis, programming (Python/Go/C/C++/Java), abstracting complex security problems into evaluable tasks
Preferred skills
CVE/CTF/real-world攻防 experience, CTF challenge/benchmark/fuzzing/security evaluation environment construction, LLM Agent/Benchmark/Eval/Verifier experience
Technologies
Python, Go, C, C++, Java
Responsibilities
Design and refine cybersecurity capabilities and evaluation systems for LLMs, construct training tasks and datasets from real-world vulnerabilities, analyze model execution trajectories to identify failures, review external expert tasks and results, collaborate with model training and RL teams to iterate security capabilities
