About the job
About 10a Labs: At the forefront of AI safety and threat intelligence, 10a Labs is the trusted partner for pioneering AI laboratories, innovative unicorns, Fortune 10 enterprises, and top-tier global technology platforms. Our red teaming, model evaluations, and intelligence gathering are integral in empowering engineering, safety, and security teams to proactively counteract emerging threats and ensure the safe deployment of AI systems.
Role Overview:
- Create and execute adversarial test suites—both manual and automated—for large language models and visual content models.
- Design multilingual prompts, jailbreaking techniques, and escalation chains to explore policy edge cases. Analyze results, prioritize failures, and produce clear vulnerability reports.
- Enhance internal tools such as prompt libraries, scenario generators, and dashboards to streamline processes.
Ideal Candidate:
- Possesses 2-4 years of experience in red teaming, security research, trust & safety, or related disciplines.
- Proficient in basic scripting for tests using Python, Bash, or comparable languages, and comfortable utilizing Jupyter or prompt-engineering tools.
- Exhibits excellent communication skills in English and at least one other language relevant to global threat landscapes.
- Adopts an adversarial mindset, documents findings succinctly, and is quick to iterate.
Essential Qualifications:
- Bachelor’s degree or equivalent experience in Computer Science, Data Science, Linguistics, International Studies, or Security.
- Basic familiarity with Python and command-line utilities.
- Demonstrated enthusiasm for AI safety, adversarial machine learning, or abuse detection.
- Strong writing capabilities for producing concise vulnerability reports and comprehensive analyses.
- Ability to swiftly adapt across various domains, modalities, and areas of abuse.
- Passionate about working in a dynamic and ambiguous environment.
Preferred Qualifications:
- Full professional proficiency in Arabic, Chinese, Farsi, Portuguese, Russian, or Spanish, alongside English.
- Previous experience in content moderation, disinformation analysis, or cyber-threat intelligence.
- Familiarity with prompt automation frameworks (e.g., Promptfoo, LangChain, Garak) and vector search or LLM fine-tuning processes.
- Formal training or certification in red teaming or penetration testing.
Compensation & Benefits:
- Salary range: $70K–$90K depending on experience and qualifications.