Principal Scientist, AI Agents & WorkSpaces Strategy

Socket.dev; Sunnyvale (CA); On-site; USD 228,000 - 309,000; Full time

Job summary

Amazon's AWS Applied AI Solutions seeks a Principal Applied Scientist to own the scientific vision across Agentic WorkSpaces. You will define how we measure and guarantee performance of AI agents and human-AI teams at enterprise scale.

You will set the science agenda, frame ambitious research directions, and mentor senior scientists while delivering measurable impact for customers and partners. This is a high-visibility role shaping the future of secure agent-computer interaction.

Qualifications

• PhD in Computer Science, Machine Learning, or a related field.
• 10+ years of applied science experience, including ML at scale.
• Experience setting research direction across multiple teams.
• Deep expertise in modern ML, including LLMs, evaluation, and experimental design.

Responsibilities

• Set the long-term scientific vision for agent performance, evaluation, and learning across Agentic WorkSpaces.
• Solve highly ambiguous, novel problems in agent perception, reasoning, and human-AI collaboration.
• Own the evaluation and measurement foundation with benchmarks, datasets, and metrics.
• Drive cross-organizational scientific alignment across partner teams and product portfolio.
• Deliver measurable business impact through improved accuracy, cost, and productivity.
• Raise the scientific bar by mentoring senior scientists and enforcing rigorous experimentation.
• Advance state of the art through publications, patents, and open-source contributions.

Job description

Amazon's AWS Applied AI Solutions seeks a Principal Applied Scientist to own the scientific vision across Agentic WorkSpaces. You will define how we measure and guarantee performance of AI agents and human-AI teams at enterprise scale.

You will set the science agenda, frame ambitious research directions, and mentor senior scientists while delivering measurable impact for customers and partners. This is a high-visibility role shaping the future of secure agent-computer interaction.