Security AI Research Engineer — Introspection Probes

Google LLC; Mountain View, New York (CA, NY); On-site; USD 174,000 - 252,000; Full time

Job summary

Google DeepMind seeks a Research Engineer for the Introspection Probe in the Gemini Security team. You will work on large-scale LLM training, serving infrastructure, and developing security/safety classifiers to protect models and users.

As part of a research-focused software engineering role, you will implement real-time interventions and collaborate with cross-functional teams across Google/DeepMind. Location preference includes Mountain View, CA, or remote options in the US.

Qualifications

• Bachelor's degree in Computer Science or related field, or equivalent practical experience.
• Experience with large-scale LLM training and serving infrastructure.
• Experience training security or safety classifiers for large AI models.
• Experience deploying security or safety classifiers.

Responsibilities

• Research, build and land improvements (training and inference) for Gemini security introspection heads.
• Research, build and land real-time interventions triggered by introspection heads.
• Partner with introspection team to improve heads training, serving and deployments.

Skills

LLM training

LLM serving infra

Security classifier training

Classifier deployment

Real-time interventions

Communication

Education

BSc in CS or related field

Job description

Google DeepMind seeks a Research Engineer for the Introspection Probe in the Gemini Security team. You will work on large-scale LLM training, serving infrastructure, and developing security/safety classifiers to protect models and users.

As part of a research-focused software engineering role, you will implement real-time interventions and collaborate with cross-functional teams across Google/DeepMind. Location preference includes Mountain View, CA, or remote options in the US.