This list contains only the countries for which job offers have been published in the selected language (e.g., in the French version, only job offers written in French are displayed, and in the English version, only those in English).
We are building a frontier AI team inside the Office Product Group to push the next generation of agentic AI systems embedded directly into Office applications. This team sits at the intersection of research and product: inventing new model- and system-level capabilities, and shipping them at global scale to hundreds of millions of users. As an Applied Scientist, you will design, train, evaluate, and deploy agentic AI systems that reason, act, and collaborate across tools, documents, and other agents—grounded in real product constraints.
Job Responsibility:
Design and build agentic AI systems for Office apps (Excel, Word, PowerPoint, Outlook), including planning, tool use, memory, and long-horizon task execution
Develop agent-to-agent communication frameworks, enabling coordination, delegation, and verification across specialized agents
Advance low-resource code generation and transformation, including adapting models to domain-specific languages, formulas, and semi-structured representations
Fine-tune and post-train foundation models (SFT, preference optimization, RLHF/PO variants) for reliability, controllability, and product-grade behavior
Own evaluation and benchmarking: design task-level, system-level, and human-in-the-loop evals that reflect real Office workflows—not just academic metrics
Partner closely with engineers and PMs to turn research ideas into scalable, shippable features with clear UX and performance guarantees
Requirements:
Bachelor's Degree in Statistics, Econometrics, Computer Science, Electrical or Computer Engineering, or related field AND 2+ years related experience (e.g., statistics, predictive analytics, research)
OR Master's Degree in Statistics, Econometrics, Computer Science, Electrical or Computer Engineering, or related field AND 1+ year(s) related experience (e.g., statistics, predictive analytics, research)
OR Doctorate in Statistics, Econometrics, Computer Science, Electrical or Computer Engineering, or related field
OR equivalent experience
Nice to have:
Understanding of evaluation methodology for generative and agentic systems, including offline benchmarks and online experimentation
Hands-on experience with LLMs and agentic systems, including prompting, tool calling, memory, and orchestration
Experience with fine-tuning and post-training large models (SFT, preference learning, reward modeling, or RL-style approaches)
Ability to move comfortably between research prototypes and production systems
Communication skills and a product mindset—comfort owning ambiguous problems end-to-end