Research Engineer, Personalization & AI Agent Orchestration

DeepMind Technologies Limited; Mountain View (CA); On-site; USD 174,000 - 252,000; Full time

Job summary

Google DeepMind seeks researchers with a strong foundation in evaluating AI systems and deploying agented frameworks. You will contribute to large-scale experiments, prototype architectures, and collaboration across research programs to push the boundaries of AI capabilities.

Focus areas include RL, SFT, and personalization techniques, along with building production-grade agents that effectively manage tools and state. Expect an impact-driven role with broad opportunities in safety and ethics.

Qualifications

• Bachelor’s degree or equivalent practical experience.
• 5 years designing, constructing, and deploying evaluation frameworks that capture programmatic or algorithmic performance.
• 1 year of agentic memory research experience.
• Experience training and fine-tuning models in environments with accelerators, with a track record of deploying production-grade AI agents that manage tool-use and complex state management.

Responsibilities

• Drive research on post-training techniques (e.g., RL, SFT, and preference optimization) specifically tailored for personalization scenarios.
• Develop novel evaluation frameworks to measure model quality against user behaviors/feedback.
• Design and train agents capable of orchestrating tools and APIs to deliver hyper-personalized experiences.

Skills

Evaluation frameworks design

Agentic memory research

Production-grade AI agents deployment

Tool-use and state management

Education

Bachelor’s degree or equivalent practical experience

Job description

Google DeepMind seeks researchers with a strong foundation in evaluating AI systems and deploying agented frameworks. You will contribute to large-scale experiments, prototype architectures, and collaboration across research programs to push the boundaries of AI capabilities.

Focus areas include RL, SFT, and personalization techniques, along with building production-grade agents that effectively manage tools and state. Expect an impact-driven role with broad opportunities in safety and ethics.