Sr Research Scientist

Microsoft Microsoft · Big Tech · Shanghai, Shanghai, China · Research Sciences

Senior Research Scientist to lead cutting-edge projects from concept to product, focusing on AI for Developers. This role involves building and training state-of-the-art models, applying and advancing techniques for leveraging LLMs in software engineering (including RAG and evaluation), and collaborating with product teams to run large-scale experiments and improve AI solutions. The primary focus is on AI for code completion and editing, retrieval-augmented systems for codebases, and efficient inference algorithms for code generation.

What you'd actually do

  1. Lead the development and evaluation of state-of-the-art models for code completion and editing, pushing the boundaries of code understanding, generation, fix and review.
  2. Develop retrieval-augmented systems that improve a model’s awareness of large and complex codebases, enabling context-rich code assistance.
  3. Design and prototype efficient inference algorithms to enable fast, interactive code generation experiences at scale.
  4. Collaborate across disciplines with product teams across Microsoft and Github
  5. Stay up to date with the research literature and product advances in AI for software engineering

Skills

Required

  • 4-year degree in computer science or engineering AND 3+ years related-research experience OR equivalent experience
  • Strong academic work and professional experience in statistics, machine learning, including deep learning, NLP, econometrics
  • Experience with LLMs in natural language, AI for code or related fields
  • Experience in productizing AI and collaborating with multidisciplinary teams

Nice to have

  • PhD degree in Computer Science, Statistics, Physics/Math, or related fields
  • Experience in building cloud-scale systems and experience working with open-source stacks for data processing and data science is desirable
  • Experience in productizing machine learning models is desirable
  • Excellent communication skills, ability to write and present research papers
  • Strong teamwork and collaboration skills

What the JD emphasized

  • state-of-the-art models
  • LLMs
  • RAG
  • evaluation
  • large scale experiments
  • code completion
  • code editing
  • code understanding
  • code generation
  • code review
  • retrieval-augmented systems
  • complex codebases
  • code assistance
  • efficient inference algorithms
  • fast, interactive code generation experiences

Other signals

  • AI for Developers
  • LLMs in software engineering
  • AI for code advancements