Principal Applied Scientist (AI)

TL;DR

Principal Applied Scientist (AI): Developing core agentic systems for M365 Copilot with an accent on reasoning, memory management, and agent orchestration at enterprise scale. Focus on translating state-of-the-art research into production-ready features, defining scientific quality bars, and driving technical excellence within the AI team.

Location: Must be based in Beijing, China

Company

A global technology leader empowering every person and every organization on the planet to achieve more through innovative AI and productivity software.

What you will do

• Design and ship advanced agentic AI techniques, including planning, reasoning, memory, and LLM post-training.
• Translate cutting-edge research into scalable, reliable, and measurable production solutions.
• Define quality bars and evaluation strategies for agent behavior, reasoning, and safety.
• Diagnose system failure modes and drive systematic, data-driven improvements.
• Provide technical leadership, mentor team members, and influence overall system architecture.

Requirements

• Bachelor’s degree and 6+ years of experience, Master’s and 4+ years, or Doctorate and 3+ years in a relevant technical field.
• Strong algorithmic and modeling skills with proven problem-solving capabilities.
• Demonstrated expertise in multi-turn agent orchestration and LLM lifecycle.
• Excellent communication skills for cross-functional collaboration.
• Must be able to work on-site in Beijing, China.

Nice to have

• Academic background or research focus in reinforcement learning (RL), agent systems, or search systems.
• Experience with long-horizon tasks and context management.

Culture & Benefits

• Opportunity to work on enterprise-grade AI products impacting millions of users.
• Collaborative environment with access to massive-scale production data.
• Strong culture of technical mentorship and scientific rigor.
• Commitment to equal opportunity and workplace inclusivity.