Research Intern - Onedrive and Sharepoint (summer 2026)

Microsoft Microsoft · Big Tech · Redmond, WA +1 · Applied Sciences

Research intern to investigate, propose, implement, and evaluate new approaches in LLMs, Multimodal AI, Reinforcement Learning, Conversational AI, and AI Agents for OneDrive and SharePoint. Will conduct experiments, develop AI models, design evaluation metrics, build datasets, and deliver models for content understanding and use across various modalities. Focus on applying scientific methods to real-world problems in content and collaboration.

What you'd actually do

  1. Conduct experiments and develop novel AI models and algorithms to address complex ODSP scenarios (e.g., intelligent document understanding, search/RAG, recommendation, generative experiences, proactive knowledge mining).
  2. Design rigorous evaluation metrics, methodologies, and validation experiments to measure performance and quality of devised AI solutions and agentic AI workflows.
  3. Build datasets (including leveraging privacy-preserving synthetic data techniques) to fine tune and benchmark models for ODSP applications.
  4. Deliver models and algorithms for content understanding, enrichment, and use at scale, across a range of different modalities, including text, images, and video.
  5. Present research findings and propose ideas in team discussions; share results through documentation and presentations and contribute to research publications.

Skills

Required

  • Currently enrolled in a PhD program in Computer Science, Electrical/Computer Engineering, Data Science, Statistics, or related fields.

Nice to have

  • Demonstrated foundation in machine learning and artificial intelligence.
  • Hands-on experience with modern deep learning techniques (e.g., transformer models, large language models).
  • Practical Python coding experience with PyTorch or similar frameworks.
  • Ability to prototype and implement algorithms efficiently.
  • Proficient analytical, problem solving, communication, and collaborative skills.
  • Able to formulate hypotheses, drive experiments, and work effectively in a collaborative environment.
  • Demonstrated research impact through publications or projects in relevant AI domains (e.g., natural language processing, information retrieval, computer vision, multimodal AI, knowledge mining).
  • A track record of applying scientific methods to real-world problems is a plus.
  • Familiarity with LLM training or fine-tuning, particularly using reinforcement learning techniques or AI agent orchestrators.
  • Experience working with large-scale datasets, enterprise content, or content management systems.
  • Experience with privacy-preserving evaluation methods or synthetic data generation is highly desirable.

What the JD emphasized

  • PhD program
  • novel AI models and algorithms
  • agentic AI workflows
  • privacy-preserving synthetic data techniques
  • content understanding
  • multimodal
  • publications at leading academic venues

Other signals

  • LLMs
  • Multimodal AI
  • Reinforcement Learning
  • AI Agents
  • large-scale datasets
  • content understanding
  • generative experiences