Understand the implications of new foundational research breakthroughs—across model architectures, multimodality, on-device efficiency, reinforcement learning, interactive environments, and trainable agent harnesses—on our open-weight model family (Gemma variants) and the broader AI ecosystem. Build collaborations with research teams, core model development teams, and product teams to ship open-weight model variants, post-training and mid-training environment runtimes, and modular agent harnesses—creating a two-way flywheel where open-source ecosystem innovations directly accelerate internal model training and product capabilities.
Work with Google product areas and external partners to understand customer and societal needs that are best met through customizable open-weight models, on-device deployment, and zero-data-leakage evaluation and regulatory sandboxes. Build a thriving open ecosystem of developer tooling, standardized environment SDKs, formal verification and safety guardrails, and day-one integrations with popular open-source frameworks, agent protocols, and inference runtimes.
Job description provided by Google.
View the original posting
