Company hidden предлагает значимую исследовательскую работу по безопасности ИИ с четким фокусом на постобучении Gemini, взаимодействием с разными командами и сильной исследовательской средой.
Актуальная область
Понятные задачи
Сильная исследовательская среда
Мэтч & Сопровод
Для мэтча с этой вакансией нужен Plus
Подключить Plus
Описание вакансии
Текст:
Hirify AI/Оригинал
TL;DR
Research Scientist, Gemini Agent Safety Post Training, DeepMind (AI): Developing safety protocols, alignment methods, and post-training techniques for Gemini and generative AI models with an accent on model behavior evaluation, safety mitigation, and large-scale SFT and RL. Focus on identifying safety gaps, optimizing models under tight launch timelines, and coordinating research, evaluation, post-training, and product teams to deploy protections.
Location: Mountain View, California, USA
Salary: $174,000–$252,000 USD annually, plus 15% bonus target, equity, and benefits.
Company
hirify.global is an AI research lab developing artificial intelligence technologies for scientific discovery, public benefit, and high-quality products.
What you will do
Design, implement, and maintain protocols for identifying safety gaps in Gemini and generative AI model behavior.
Feed safety insights into post-training to improve in-model and out-of-model protections.
Develop and apply modeling and alignment techniques for real-world safety challenges under tight launch timelines.
Advance model optimization through Supervised Fine-Tuning and Reinforcement Learning at massive scale.
Collaborate with post-training, evaluation, and product teams to align incentives, close safety gaps, and influence safety decisions.
Requirements
PhD in Computer Science, a related field, or equivalent practical experience.
At least 1 year of experience with Generative AI, Large Language Models, natural language processing, or agent-based systems.
Nice to have
Experience in AI safety and model alignment.
Experience developing AI agents, safety-critical systems, or machine learning infrastructure.
Expertise in reward modeling, Reinforcement Learning for LLMs, instruction tuning, and long-range RL.
Ability to take research concepts through product realization, with strong engineering and experimental experience.
Publications at AI or ML venues such as NeurIPS, ICLR, ICML, EMNLP, AAAI, or UAI.
Culture & Benefits
Collaborative internal culture centered on mutual dedication and cooperation.
Coordination with interdisciplinary research, post-training, evaluation, and product teams.
15% 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 →