Principal Applied Scientist

Microsoft • Redmond, WA

76d

About The Position

The MSAI Search Relevance team is at the forefront of delivering world-class search quality across Microsoft’s ecosystem. We are the driving force behind the relevance of results in Copilot Search experiences and serve as the core retrieval layer in the RAG architecture powering Bizchat. Our impact also extends to maintaining high search quality across traditional endpoints like Outlook, Teams, and SharePoint Search. Our team thrives at the intersection of innovation and applied machine learning. We are looking for a Principal Applied Scientist to help us deliver breakthrough applied machine learning and information retrieval solutions at enterprise scale. This role is a unique opportunity to apply state-of-the-art techniques—including dense retrieval, hybrid search, multilingual large language models (LLMs), RAG (Retrieval-Augmented Generation), and transformer-based re-ranking models and agentic search—to solve complex challenges in Copilot-driven enterprise search.

Requirements

• Strong background in applied machine learning and information retrieval.
• Experience with dense retrieval, hybrid search, and multilingual large language models (LLMs).
• Proficiency in advanced vector search algorithms and semantic retrieval.
• Expertise in deep language understanding and relevance model design.
• Ability to build evaluation pipelines and conduct A/B testing.
• Strong collaboration skills to work across different teams and time zones.

Nice To Haves

• Experience in AI-powered search technologies.
• Familiarity with transformer-based models and agentic search.
• Previous experience in a leadership or strategic role.

Responsibilities

• Deliver mission-critical innovations that directly improve Copilot experiences such as Agentic Search in modern orchestrator architectures.
• Adapt advanced vector search algorithms (e.g., FAISS, ANN, ScaNN) for enterprise-scale semantic retrieval.
• Improve classic and neural keyword search quality through deep language understanding.
• Design and train relevance models, including LLM fine-tuning and learning-to-rank (LTR) approaches.
• Build robust evaluation pipelines using offline metrics and online A/B experimentation.
• Drive cross-org collaboration with platform partners, other applied science teams, and product teams across time zones.

What This Job Offers

• Job Type: Full-time
• Career Level: Principal
• Number of Employees: 5,001-10,000 employees