Gemini Post-training Software Engineer, DeepMind

Overview In this role you will contribute to Gemini by advancing digital agents and deploying them into production. You will identify model improvement areas, and train and evaluate agents across diverse digital environments. The focus is on multi-step reinforcement learning in multimodal settings to push the capabilities of AI agents. You will work with cross-functional teams to drive meaningful impact while upholding safety and ethics in AI. Compensation / Benefitsequitybonus targetbenefits ResponsibilitiesMake core contributions to Gemini to advance digital agentsIdentify model improvement areas for agentsTrain and evaluate agents across diverse digital environmentsSpecialize in multi-step reinforcement learning in multimodal digital environments Key requirements5 years of experience designing, training, and evaluating ML modelsExperience in Python programmingExperience with ML test writing for TensorFlow or similar for post-silicon data collection and pre-silicon modelingExperience with Large Language Models, NLP, or Generative AIleadership qualitiescollaborationadaptabilityReinforcement LearningPython programmingTensorFlow or similar