Introspection Probe Research Engineer, Development and Infrastructure, Gemini Security, DeepMind
Overview:
In this role, you will help design and deploy improvements for Gemini security through training and real-time interventions. You'll collaborate with the introspection team to enhance training, serving, and deployment of safety classifiers for large AI models. You'll run large-scale experiments, prototype architectures, and contribute to real-world protection of Gemini from unauthorized distillation. This position blends research with practical deployment to advance AI safety at scale. Compensation / Benefits 15% bonus targetequitybenefitslocation flexibilityremote-friendly optionscareer growth opportunities Responsibilitiesresearch and implement improvements for training and inference pipelinesdesign and land real-time interventions triggered by introspection headscollaborate with the introspection team to optimize training, serving and deploymentsbuild and evaluate security/safety classifiers for large AI modelsdeploy classifiers and monitor performance in productionconduct large-scale experiments and prototyping of architectures
Key requirements:
Bachelor's degree in Computer Science or related field or equivalent practical experienceexperience with large-scale LLM training and serving infrastructureexperience training security or safety classifiers for large AI modelsexperience deploying security or safety classifiersexceptional communicationability to influence cross-functional stakeholdersstrong collaboration across teamsLLM training and serving infrastructuretraining and deploying security/safety classifiersreal-time intervention design for AI systems