Minimum qualifications:

• Bachelor's degree in Computer Science, a related technical field, or equivalent practical experience.
• 5 years of experience in software design and development using Python, distributed systems, or cloud infrastructure.
• Experience building, deploying, and operating multi-agent or AI systems in production or near-production environments.
• Experience with evaluation frameworks, metrics design, or quality measurement for machine learning systems.
• Experience collaborating with cross-functional research teams, external partners, or enterprise customers to translate requirements into technical solutions.
  

Preferred qualifications:

• Master's degree or PhD in Computer Science, Artificial Intelligence, or a related field.
• Experience with LLM agents, autonomous multi-step reasoning systems, meta-learning, or self-improving ML pipelines.
• Experience with large-scale data pipelines (e.g., Apache Beam) or foundation model training and fine-tuning at scale.
  
• Experience working in research environments or track record of published research in relevant AI/ML conferences.
  
• Domain experience in life sciences, drug discovery, cybersecurity, or developer tools/coding agents.

About the job

Google DeepMind's Science and Strategic Initiatives unit is building a new team focused on the commercialization and real-world academic impact of AI models across Science, Cybersecurity (CodeMender), and Coding Agents. We sit at the intersection of Google DeepMind's frontier research and Google Cloud's enterprise reach, transferring breakthroughs into products that generate groundbreaking discoveries and commercial impact with exceptional institutions (Harvard, Broad Institute, Roche, AstraZeneca), leading enterprises, and internal Google teams.

We are building a self-improving meta-agent framework to automatically diagnose failure modes, generalize learnings across customer engagements, and continuously improve agent quality at scale. We look for engineers comfortable building production systems and reasoning about research problems, who thrive in ambiguity, engage directly with customers, and want to see AI agents work in the real world—not just on benchmarks.

In this role, you will design, build, and operate the scaffolding and meta-agent framework across four key failure categories. You will build diagnostic agents and automated validation pipelines to detect and remediate real-world integration issues before they impact agent quality. You will design robust evaluation frameworks to measure production performance when lab benchmarks fail, and identify when evaluation methodology itself is the root cause of perceived failures. You will develop memory and knowledge architectures that extract, distill, and generalize insights across multi-agent deployments to prevent learnings from remaining episodic. Additionally, you will characterize model capability gaps with empirical evidence, partnering with Google DeepMind research teams to drive targeted model improvements.

Artificial intelligence will be one of humanity’s most transformative inventions. At Google DeepMind, we are a pioneering AI lab with exceptional interdisciplinary teams focused on advancing AI development to solve complex global challenges and accelerate high-quality product innovation for billions of users. We use our technologies for widespread public benefit and scientific discovery, ensuring safety and ethics are always our highest priority.

Responsibilities

• Design and build the self-improving meta-agent framework—scaffolding, diagnostic tools, and feedback loops across science, cybersecurity, and coding agents.
• Deploy, monitor, and improve AI agents in real-world settings with enterprise customers and academic partners, transitioning direct engagements into a scalable deployment model.
• Build automated evaluation pipelines that capture real-world agent quality, robustness, and performance beyond lab benchmarks.
• Develop cross-engagement learning systems and memory architectures that extract and generalize insights across multi-agent deployments.
• Characterize core foundation model limitations with empirical evidence, collaborating with Google DeepMind research teams to drive foundational improvements.