Staff Research Engineer, Generative AI, Cloud AI Research, Co-Scientist

Google is hiring a Staff Research Software Engineer to join its Cloud AI Research team in Zurich. The role focuses on turning scientific AI research into robust software, with a special emphasis on Co-Scientist, a multi-agent AI system built on Gemini that acts as an autonomous virtual scientific colleague. It’s a full-time, onsite position based in Zurich, ideal for engineers who want to translate academic ideas into production-ready systems and work closely with researchers, product managers, and external partners.

Co-Scientist is designed to help researchers synthesize literature, evaluate hypotheses, and optimise experimental designs by using a generate, debate, and evolve approach. The successful candidate will serve as the principal technical bridge, translating AI research into scalable software and leading the Co-Scientist efforts from the Zurich team. The role sits within Google Cloud AI Research, where the goal is to push forward AI capabilities that can impact customers across industries while collaborating with product teams to deliver real-world results.

If you enjoy tackling technical challenges at the intersection of AI research and software engineering, mentoring teammates, and shaping the architecture of complex systems, this could be a fit. The project requires versatility and leadership as it evolves with the company’s needs and tech advances.

Shaping multi-agent intelligence in Zurich

In this role you’ll own projects from their initial scientific ideas through software design, testing, deployment, and performance analysis. You’ll lead the creation of agentic AI systems, building multi-agent prompting techniques and asynchronous task execution frameworks to support scalable reasoning.

Expect to work across the stack with Python, C++, and Google’s core backend components, crafting software that is robust, maintainable, and capable of handling large-scale workloads. You will design and implement evaluation pipelines, automated test suites, and diagnostic datasets to quantify agent performance as systems grow more complex.

Collaboration is a core part of the job. You’ll partner with Google DeepMind, Google Research, and university labs to tackle difficult scientific problems, while also mentoring junior engineers and upholding engineering quality throughout the Co-Scientist initiative in Zurich.

Must-haves and nice-to-haves

• Bachelor’s degree or equivalent practical experience, with a solid foundation in software development.
• Five years or more building software, including testing and bringing products to users.
• Three years spent optimizing ML infrastructure, covering model deployment, evaluation, data processing, debugging, and fine tuning.
• Two years of experience with GenAI techniques or related concepts such as language modeling or computer vision.

• Advanced degree in computer science, applied math, computational sciences, or a closely related field (Master’s, PhD, or Postdoc).
• Three years of hands-on work with deep learning frameworks (JAX, TensorFlow, or PyTorch) and a strong algorithmic background.
• One year in a technical leadership capacity guiding teams and projects.
• Experience driving independent research-adjacent software development and turning academic ideas into scalable products.
• A track record of technical or scientific contributions to complex, large-scale projects.

Location, schedule and practical details

This is a full-time, onsite role based in Zurich.

Tips to stand out for the Co-Scientist team

Lead your resume with examples of projects you owned from concept to delivery, including how you measured success and what the results were, especially for end-to-end software work tied to AI or ML systems.

Back up the must-have skills with concrete outcomes. Describe specific deployments you led, how you improved model evaluation or data processing, and how you debugged and fine-tuned models in production settings.

Prepare to discuss multi-agent AI and agent prompting in depth. Have a clear story about a project where you designed or improved asynchronous task execution and how you tested these systems at scale, including any challenges you faced and how you addressed them.

Think of a thoughtful, specific question to ask about Co-Scientist’s roadmap or collaboration patterns. For example, you could ask how the Zurich team collaborates with Google DeepMind and Google Research, or how Gemini capabilities are evolving to support autonomous scientific reasoning.