Senior Staff Research Scientist, Gemini Safety Post-Training, DeepMind
DeepMind – Mountain View, CA, USA
Required qualifications
* PhD in Computer Science, a related field, or equivalent practical experience.
* 6 years of experience in Machine Learning Algorithms and Language Modeling.
* One or more scientific publications in the ML/AI conferences or journals (e.g., NeurIPS, ICML, ICLR, CVPR).
Preferred qualifications
* 5 years of experience in safety/alignment, including RLHF, reward modeling, and out-of-model safety systems. Proven track record of mitigating model risks at scale.
* 5 years of documented experience driving research concepts from initial hypothesis through to product realization.
* Experience designing and deploying AI agents and safety-critical, high-availability systems.
* Expertise in designing/executing comprehensive model evaluation frameworks to identify, quantify, and close critical safety gaps.
* Deep technical experience across the full LLM life-cycle, including pre-training, inference optimization, and fine-tuning.
About the job
As models become more agentic, executing long-horizon tasks, using tools, writing and running code, and operating across multi-step workflows, the challenge of making them safe fundamentally changes. Surface‑level safety methods (output filtering, refusal tuning, policy guardrails) were designed for single‑turn interactions and are not enough for agents that plan, act, and adapt over extended horizons.
We are looking for a Senior Staff Research Scientist to rethink safety post‑training for this new reality. You will bring frontier post‑training expertise to develop training methods that make Gemini models deeply safe and aligned, especially in agentic settings. This role sits in Gemini Safety and partners closely with the Artificial General Intelligence (AGI) Safety team and the Gemini post‑training organization.
Responsibilities
* Rethink how safety is trained into models, especially for agentic, long‑horizon behavior.
* Design and ship post‑training recipes (Reinforcement Learning (RL), Supervised Fine‑Tuning (SFT), and beyond) that install safety and alignment properties into Gemini models. Own the path from research to production.
* Build the metrics and evaluations that tell us whether training is actually making models safer in deployment, not just on benchmarks.
* Work directly with the post‑training pipeline and infrastructure. Partner with the AGI Safety team to bring alignment research into practical training. Translate between research and production.
* Shape the roadmap for where safety post‑training goes next. Build and grow the team to execute on it.
Benefits
US: $262,000 - $365,000 (USD) + 25% bonus target + bonus + equity + benefits.
EEO Statement
Google is proud to be an equal opportunity and affirmative action employer. We are committed to building a workforce that is representative of the users we serve, creating a culture of belonging, and providing an equal employment opportunity regardless of race, creed, color, religion, gender, sexual orientation, gender identity/expression, national origin, disability, age, genetic information, veteran status, marital status, pregnancy or related condition (including breastfeeding), expecting or parents-to-be, criminal histories consistent with legal requirements, or any other basis protected by law. See also Google's EEO Policy, Know your rights: workplace discrimination is illegal, belonging at Google, and how we hire.
Recruitment
Google is a global company and, in order to facilitate efficient collaboration and communication globally, English proficiency is a requirement for all roles unless stated otherwise in the job posting.
Equity is granted exclusively and discretionarily by Alphabet Inc. on the basis of an agreement concluded between you and Alphabet Inc. Alphabet Inc. is your sole contractual partner with respect to equity grants. GSU grants are not guaranteed, are discretionary, are subject to approval by the Alphabet Inc. board of directors or its delegate, the terms of the relevant Alphabet Inc. stock plan, and your grant agreement. They have no impact on statutory payments. Current or past grants do not confer an acquired right.
#J-18808-Ljbffr