Drive an ambitious research agenda in post-training, reinforcement learning, and agentic modeling to improve Gemini's reasoning, tool use, factuality and multimodal performance across global languages, dialects, and locales. Develop scalable algorithms for cross-lingual capability transfer, synthetic data generation, reward modeling, and self-improvement flywheels that close the quality gap for both major global languages and low-resource languages.
Improve local and cultural grounding, factuality, naturalness, and language consistency across long-horizon multi-turn and multimodal interactions (text, speech, image, and video). Design trustworthy evaluation methodologies, evaluations, and human-grounded reward signals to measure and steer complex agentic and culturally nuanced model behaviors. Partner closely with core Gemini post-training, agentic, and pretraining/mid-training teams—as well as product partners across Google—to land research breakthroughs directly into flagship Gemini models.
Job description provided by Google.
View the original posting
