🇯🇵 Research Scientist, Multimodal Assistive Agents (Astra) - Tokyo Full time

You will conduct cutting-edge research on large language models (LLMs), particularly in the multimodal domain (vision, audio, text), focusing on the development of more capable, robust, and helpful assistive agents, with a direct path to impacting millions of users through Google products. This role offers a unique opportunity to contribute to fundamental research in AI while collaborating with a world-class team at Google DeepMind and seeing your work translated into real-world applications. If you are passionate about shaping the future of human-computer interaction through AI and are eager to make a significant impact in the rapidly evolving landscape of assistive technologies, we encourage you to apply.

- Conduct cutting-edge research: Develop novel algorithms, models, and techniques for multimodal assistive agents, pushing the boundaries of AI research in areas such as natural language understanding, computer vision, speech processing, and reinforcement learning. Your research will directly contribute to advancing Google’s assistive agent capabilities in products.
- Develop and evaluate models: Design, implement, and evaluate multimodal assistive AI agents. Explore techniques like prompt engineering, few-shot learning, post-training techniques to improve model performance and robustness in diverse real-world scenarios. Your research will be focused on building assistive agents that can perceive, reason, plan, and interact with humans in more natural and intuitive ways, ultimately shaping user experience in Google products.
- Collaborate with a world-class team: Work closely with other research scientists, engineers, and product teams across Google DeepMind, fostering a collaborative and intellectually stimulating environment. Share your research findings through publications in top-tier conferences and journals, while also contributing to the development of impactful products.
- Contribute to real-world impact: See your research contribute to the development of next-generation multimodal assistive agents with applications across various domains, including education, healthcare, gaming, accessibility, and more, directly influencing the future of Google products and services.
- Stay at the forefront of AI research: Continuously explore emerging trends and new research directions in multimodal AI. Participate in international conferences and workshops to share your work and learn from others in the field, bringing these cutting-edge advancements to Google’s product landscape.

- Ph.D. in Computer Science, Artificial Intelligence, or a related field.
- Strong publication record in top-tier machine learning conferences or journals.
- Solid understanding of deep learning, natural language processing, computer vision, and/or speech processing.

In addition, the following would be an advantage:
- Experience with multimodal learning, large language models, and/or assistive AI agents.
- Experience with prompt engineering, few-shot learning, post-training techniques, and evaluations.
- Familiarity with large-scale model training and deployment.
- Strong programming skills in Python or similar languages.
- Excellent communication and collaboration skills.
- Japanese language skills are a plus.

TBD