Virologist – AI Safety and Evaluations (AI)
Мэтч & Сопровод
Для мэтча с этой вакансией нужен Plus
Описание вакансии
Location: Remote – United States. Candidates based in Washington, DC, New York City, the San Francisco Bay Area, or Denver are expected to work from a office at least two days per week. All new hires must complete in-person onboarding in Washington, DC, New York City, or San Francisco during the first week; occasional travel may be required.
Company
provides adversarial red teaming, model evaluations, and threat intelligence for frontier AI labs, AI companies, Fortune 10 companies, and global technology platforms.
What you will do
- Design scientific benchmarks, evaluation methodologies, scenarios, questions, and tasks for assessing advanced AI models across virology, microbiology, and related biological domains.
- Conduct model capability evaluations and red-teaming exercises focused on biological misuse risks, dual-use research, and emerging AI capabilities.
- Assess AI-generated scientific outputs for accuracy, feasibility, completeness, technical limitations, and biosecurity implications.
- Develop evaluation criteria, scoring rubrics, automated benchmarks, and scalable assessment approaches with scientists, AI researchers, engineers, and subject matter experts.
- Analyze evaluation results, identify capability and limitation patterns, and contribute to technical reports and client deliverables.
- Track developments in biological research, AI capabilities, biosecurity risks, and AI safety frameworks.
Requirements
- PhD in Virology, Microbiology, Molecular Biology, or a closely related biological science, or equivalent practical experience.
- Research experience involving pathogenic viruses, bacteria, or other biologically relevant systems.
- Strong understanding of biological research methods, experimental design, and technical challenges in biological research and development.
- Knowledge of biosecurity principles, dual-use biological research, and potential biological misuse risks.
- Experience critically evaluating scientific literature, experimental methodologies, and technical research outputs.
- Strong analytical skills for assessing complex scientific information for accuracy, feasibility, and technical limitations.
- Experience preparing technical reports, presenting scientific findings, and communicating complex concepts to multidisciplinary audiences.
Nice to have
- Experience with AI model evaluations, scientific benchmarks, LLM capability assessment, AI red-teaming, biological risk assessments, or emerging-technology misuse scenarios.
- Hands-on virology or microbiology laboratory experience, including BSL-2 or higher containment environments.
- Experience with command-line tools, Python, or computational approaches relevant to biological research.
- Familiarity with AI-assisted biological research tools, bioinformatics, scientific reasoning, AI safety frameworks, and biological or chemical risk assessment.
- Understanding of biological weapons threats, dual-use research concerns, biosecurity risk mitigation, relevant policies, international agreements, and the Biological Weapons Convention.
- Experience developing evaluation datasets, scoring rubrics, structured scientific assessment methodologies, or AI-supported biological research workflows.
- Interest in emerging AI capabilities and their implications for scientific research, biological security, and AI safety.
Culture & Benefits
- Contract engagement on a per-needed basis with commitment dependent on workload.
- Remote work is available from the United States; office-based hybrid work is expected for candidates in selected metropolitan areas.
- In-person onboarding during the first week and occasional travel for interviews, meetings, and other business needs.
- Cross-disciplinary collaboration with scientists, AI researchers, engineers, and subject matter experts.
Будьте осторожны
Если работодатель просит войти в их систему, используя iCloud/Google, прислать код/пароль, запустить код/ПО, не делайте этого. Это мошенники. Обязательно жмите «Пожаловаться» или пишите в поддержку. Подробнее в гайде →