Google DeepMind предлагает руководящую роль в исследованиях по безопасности ИИ с заметным влиянием и сильной научной базой; из-за высокого уровня это специализированная возможность.
?Команда в управлении
Безопасность ИИ
Современное ML
Чёткая зона ответственности
Мэтч & Сопровод
Для мэтча с этой вакансией нужен Plus
Подключить Plus
Описание вакансии
Текст:
Hirify AI/Оригинал
TL;DR
Research Scientist/Engineer (AI Safety) (LLMs, Multimodal Foundation Models, Reinforcement Learning): Defining safety intervention architectures and advancing post-training methods for multimodal foundation models with an accent on alignment, controllability, and high-quality generation. Focus on shaping model behavior, distilling serving-time classifiers into training rewards, and leading a team of 10 or more researchers and engineers.
Location: Mountain View, CA, USA or New York, NY, USA
Salary: $262,000–$364,000 USD annually, plus 25% bonus target, equity, and benefits.
Company
Google DeepMind is an AI research lab developing advanced technologies for scientific discovery, public benefit, and high-quality product innovation, with safety and ethics as priorities.
What you will do
Define architectural strategies for placing safety interventions across multimodal foundation models, balancing shared and bespoke approaches.
Advance generalizable post-training methods across media types as prompting and conditioning capabilities evolve.
Shape intrinsic model behavior to refuse severe harms while handling borderline requests with high generation quality and controllability.
Translate serving-time classifiers into training rewards with cross-functional teams and expand the safety–quality trade-off frontier.
Manage a team of 10 or more researchers and engineers while fostering scientific excellence and collaboration.
Requirements
PhD or equivalent practical research experience.
At least 5 years in a technical leadership role setting direction and guiding project workstreams.
At least 5 years in people management, supervision, or team leadership.
At least 5 years of machine learning experience with large language models, multimodal foundation models, or reinforcement learning, specializing in safety and alignment such as model alignment, AI safety, supervised fine-tuning, or reward modeling.
At least one peer-reviewed scientific publication or public preprint, including venues such as NeurIPS, ICML, ICLR, CVPR, ACL, or CoRL.
Nice to have
At least 6 years initiating, scoping, and executing independent research agendas or leading collaborative academic projects.
Experience with LLMs, post-training, reasoning and reinforcement learning, agents, multimodal perception or vision, robotics, audio or speech, safety and alignment, or scientific AI.
Experience bridging research breakthroughs with product engineering or external scientific and policy stakeholders.
Sustained first-author and senior-author publication record in premier conferences or journals such as NeurIPS, ICML, ICLR, CVPR, SIGGRAPH, CoRL, Science, Nature, or PNAS.
Culture & Benefits
Interdisciplinary, global research teams focused on collective excellence and complex global challenges.
Work environment emphasizing scientific discovery, public benefit, safety, and ethics.
Annual salary includes a 25% bonus target, equity, and Google benefits.
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 →