Product Manager, Science Agents, Deepmind

Google Google · Big Tech · London, United Kingdom

Product Manager role focused on developing and shipping AI systems, specifically science agents, that accelerate scientific discovery. The role involves bridging frontier research with product development, identifying impactful applications, and incubating new agentic capabilities. It requires technical depth, experience with LLMs, and familiarity with agent evaluation frameworks.

What you'd actually do

  1. Advocate for the user and become a domain expert (e.g., familiarity with science tools and their technical foundations, capabilities, and limitations).
  2. Build prototypes and partner with teams across Google to land our research in scaled products and partner with research teams to improve the way we measure and hill climb quality.
  3. Identify nascent research that could become transformative products.
  4. Flex to other initiatives in science as needed.
  5. Identify impactful applications of flagship agents to new scientific domains and help incubate new agentic capabilities to assist with real-world scientific workflows.

Skills

Required

  • Product management
  • Advanced AI/ML concepts
  • Large language models (LLMs) research and products

Nice to have

  • MBA or advanced degree in a relevant technical or business field
  • Model and agent evaluation frameworks
  • Products geared toward a technical user base
  • Evaluating the applications, risks, and benefits of new technologies
  • Ability to communicate with impact, fostering a collaborative environment that inspires and motivates cross-functional teams to achieve ambitious goals.

What the JD emphasized

  • Experience with advanced AI/ML concepts, including large language models (LLMs) research and products.
  • Experience with model and agent evaluation frameworks.

Other signals

  • Develop AI systems to accelerate science
  • Build and apply new agentic systems
  • Bridge frontier research with real-world scientific issues
  • Identify impactful applications of flagship agents
  • Incubate new agentic capabilities