Research Engineer, Astra Evals, DeepMind

Requirements:

• Bachelor's degree in Computer Science, Machine Learning, a related technical field, or equivalent practical experience.
• 5 years of experience in Python and one or more general-purpose programming languages (C++, Java, Go).
• Experience building or evaluating agentic systems using orchestration frameworks (e.g., LangChain, AutoGen, CrewAI, Semantic Kernel) or custom Large Language Models (LLMs)-based control loops.
• Experience developing machine learning pipelines.

Nice to haves:

• Master's degree or PhD in Computer Science, Machine Learning, or a related field.
• Experience with and adoption of capabilities for AI-assisted coding tools.
• Experience publishing at conferences on the topic of agentic AI and multimodal large language models (LLMs).
• Experience with model evaluation frameworks and data analysis tools.
• Excellent problem-solving skills applicable to agentic AI workflows.

What you'll be doing:

• Evaluate frontier agentic capabilities for AI assistance.
• Analyze evaluation results.
• Set and steer the reward function in reinforcement learning to improve Google's agentic capabilities.
• Transform friction points within agentic journeys into trainable examples.
• Provide insights to model and product teams.

Perks and Benefits:

• Opportunity to work at a pioneering AI lab with interdisciplinary teams.
• Engage in advancing AI development to solve complex global challenges.
• Contribute to scientific discovery and ensure safety and ethics are top priorities.
• Learning opportunities and varied career pathways for exceptional results.