research
A selection of research programs and systems I have helped lead or build.
2025–presentGemini OmniMultimodal generation and conversational editing grounded in Gemini's world understanding. 2024–2025VeoGoogle DeepMind's flagship video generation models, including Veo 1, 2, and 3. 2021–2022GatoA single generalist agent capable of more than 600 tasks across modalities and embodiments. 2017–2021Distributed reinforcement learningResearch and open-source infrastructure that helped establish actor–learner architectures at scale.