Design and roll out the next-generation multimodal and encoder architecture across workhorse models, integrating frame-level video embeddings in partnership with Google DeepMind. Lead the evolution of Content Safety’s teacher stack by migrating to next-generation Gemini models and establishing In-Context Learning Video. Lead the design of Policy enforcement RAG (PeRAG), a low-latency, high-precision retrieval system backed by a unified Long-Term Memory service. Lead the design of a greenfield Model Debug and Gap Mitigation Agent to automatically diagnose teacher-student distillation gaps. Provide technical leadership on high-impact projects.
Job description provided by Google.
View the original posting
