Faculty mudofaa, sog‘liqni saqlash, moliyaviy xizmatlar, davlat xizmatlari va chakana savdo sohalaridagi murakkab operatsion muammolarni hal etishga mo‘ljallangan ilg‘or sun’iy intellekt va mashinaviy o‘rganish yechimlarini ishlab chiqadi. Kompaniya tashkilotlar bilan hamkorlikda ilg‘or sun’iy intellektni milliy xavfsizlik, natijalarni prognozlash va jarayonlarni takomillashtirish kabi yo‘nalishlarda qo‘llaydi hamda amaliy ta’sir, mas’uliyatli joriy etish va barqarorlikka alohida e’tibor qaratadi.
Senior AI Safety Research Scientist – Evaluations at Faculty (Hybrid, UK)
Lead research into AI safety evaluations for CBRN and cyber risks at Faculty. Develop methods, tools, and evidence that support responsible deployment of advanced AI systems.
Tavsif
- Develop advanced safety evaluations for high-impact areas including CBRN and cyber.
- Conduct original AI safety research from initial concept through publication.
- Identify strategic opportunities that strengthen the organisation’s safety evaluation research agenda.
- Provide technical expertise for client projects, evaluation programmes, and frontier-lab red-teaming.
- Represent Faculty in discussions with researchers, frontier AI labs, and government stakeholders.
- Create publications, technical reports, evaluation frameworks, and research tooling.
- Work closely with Faculty’s AI safety and delivery teams.
Talablar
- Demonstrated ownership of research from identifying original questions through publication.
- Practical experience designing and implementing AI evaluations or benchmarks.
- Ability to assess construct validity and address confounding factors.
- Expertise in red-teaming, adversarial testing, jailbreaking, indirect prompt injection, and model safeguard robustness.
- Strong grounding in experimental design, statistical analysis, uncertainty quantification, and Bayesian methods.
- Thorough understanding of language models, generative AI architectures, training approaches, and safety mitigations.
- Proficiency in Python.
- Commitment to robust, reproducible research engineering practices.
- Experience with threat and risk modelling is desirable.
- Knowledge or experience in CBRN or cybersecurity is desirable.
- Record of influential AI research demonstrated through leading publications or comparable practical work is desirable.
- UK visa sponsorship requirements are assessed during the application process.
Imtiyozlar
- Part-time working arrangements can be discussed.
- Recruitment approach centred on diversity and inclusion.
- Access to AI tools for research and interview preparation.
- Every application receives human review.
- Option to opt out of the AI interview note-taker.