Research Scientist, Gemini Safety

Google DeepMind; Mountain View (CA); On-site; USD 130,000 - 160,000; Full time

Job summary

A leading AI research company is seeking a Research Scientist focused on enhancing safety and fairness in AI models. Responsibilities include tuning state-of-the-art LLMs and developing evaluation protocols. The ideal candidate will hold a PhD in Computer Science and have significant post-training experience with LLMs. Join an innovative team dedicated to advancing AI for public benefit in an inclusive environment that values diversity.

Qualifications

• Significant post-training experience with LLMs.
• Experience in Reward modeling and Reinforcement Learning for LLMs.
• Track record of publications at major AI conferences.

Responsibilities

• Post-training / instruction tuning state of the art LLMs across various modalities.
• Design and maintain evaluation protocols for model safety and fairness.
• Drive innovation in Supervised Fine Tuning and Reinforcement Learning.

Skills

Post-training experience

LLMs instruction tuning

Collaborating on applied research projects

Education

PhD in Computer Science or related field

Tools

JAX

Job description

Overview

Snapshot

Artificial Intelligence could be one of humanity’s most useful inventions. At Google DeepMind, we’re a team of scientists, engineers, machine learning experts and more, working together to advance the state of the art in artificial intelligence. We use our technologies for widespread public benefit and scientific discovery, and collaborate with others on critical challenges, ensuring safety and ethics are the highest priority.

The Role

We’re looking for a versatile Research Scientist at ease both with figuring out how to approach new research questions, and the technical implementation of research ideas. Our team focuses on advancing the safety and fairness behavior of state of the art AI models. We drive the development of the foundational technology adopted by numerous product areas including Gemini App, Cloud API, and Search.

Key responsibilities

• Post-training / instruction tuning state of the art LLMs, focusing on text-to-text, image/video/audio-to-text modalities and agentic capabilities
• Exploring data, reasoning and algorithmic solutions to make sure Gemini Models are safe, maximally helpful, and work for everyone
• Improve Gemini’s adversarial robustness, with a focus on high-stakes abuse risks
• Design and maintain high quality evaluation protocols to assess model behavior gaps and headroom related to safety and fairness
• Develop and execute experimental plans to address known gaps, or construct entirely new capabilities
• Drive innovation and enhance understanding of Supervised Fine Tuning and Reinforcement Learning fine-tuning at scale

About You

In order to set you up for success as a Research Scientist on the Gemini Safety team we look for the following skills and experience:

• PhD in Computer Science, a related field, or equivalent practical experience
• Significant LLM post-training experience

In addition, the following would be an advantage

• Experience in Reward modeling and Reinforcement Learning for LLMs Instruction tuning
• Experience with Long-range Reinforcement learning
• Experience in areas such as Safety, Fairness and Alignment
• Track record of publications at NeurIPS, ICLR, ICML, RL/DL, EMNLP, AAAI, UAI
• Experience taking research from concept to product
• Experience with collaborating or leading an applied research project
• Experience with JAX

At Google DeepMind, we value diversity of experience, knowledge, backgrounds and perspectives and harness these qualities to create extraordinary impact. We are committed to equal employment opportunity regardless of sex, race, religion or belief, ethnic or national origin, disability, age, citizenship, marital, domestic or civil partnership status, sexual orientation, gender identity, pregnancy, or related condition (including breastfeeding) or any other basis as protected by applicable law. If you have a disability or additional need that requires accommodation, please do not hesitate to let us know.