高级漏洞利用研究专家 - 安全与风控
Core
Research and exploit vulnerabilities in large models to enhance their cybersecurity capabilities and train models on complex exploitation tasks.
Role type
Senior IC security researcher (LLM cybersecurity)
Builds
Cyber-capable large language models and safety products
Domain
AI/ML security, Kernel security, Exploitation research
Deliverable
production ML models
Required skills
Kernel exploitation (Linux/Windows/OpenHarmony), Vulnerability root cause analysis, Low-level source code reading, Memory management and concurrency understanding, System call mechanisms, Debugging tools (GDB/WinDbg/QEMU/KASAN), CTF competition experience
Preferred skills
Complex CVE research, syzkaller familiarity, Cross-version vulnerability analysis, Kernel CTF question design
Technologies
GDB, WinDbg, QEMU, KASAN, syzkaller
Responsibilities
Analyze model solving trajectories to identify capability gaps, Design high-quality training tasks and expert trajectories for model cyber training, Build reproducible experimental environments and oracles for vulnerability analysis, Research and validate model reasoning and tool usage in complex exploitation scenarios, Collaborate with algorithm and engineering teams to iterate training data and agent toolchains
Seniority
Senior, hands-on IC
