Skip to main content

Research Intern Frontier Tuning

MicrosoftRedmond, WAPosted 18 hours ago

Pay
$82k – $162k
per year
Type
Internship
Summer 2027
Where
Redmond, WA
+2 more locations
Who
PhD
Students
Apply by
Dec 17, 2026
Date from the posting

Overview

Research Internships at Microsoft offer a dynamic environment for research careers, with a network of world-class research labs led by globally recognized scientists and engineers. They pursue innovation across scientific and technical disciplines, working on complex challenges in computing, healthcare, economics, and the environment.

The Frontier Tuning research team in Microsoft 365 is seeking Research Interns for summer 2027 in Redmond, WA, the San Francisco Bay Area, CA, or Baltimore, MD.

The team's mission is to help every organization own its intelligence: AI that knows, reasons, and works the way the organization does. The team captures how an enterprise works, including its knowledge, workflows, tools, and expert judgment, in reinforcement learning environments (RLEs). These environments act as a gym where AI agents improve by tuning their skills, evaluations, harnesses, and underlying models, while customer data stays within the customer's trust boundary. This research is the foundation of Frontier Tuning, announced at Microsoft Build 2026 and available to makers and developers in Microsoft Copilot Studio, Microsoft Foundry, and GitHub Copilot.

Enterprise work is a demanding testbed for AI. Tasks are long-horizon and open-ended, quality is judged against expert rubrics rather than a single correct answer, and knowledge is spread across documents, conversations, slides, and drawings. The team has shown that reinforcement fine-tuning in well-built enterprise environments lets smaller tuned models outperform the latest frontier models on such tasks, at lower cost.

Research Interns will advance the science of adapting AI agents to the enterprise. Projects may focus on topics such as:

  • Reinforcement learning and post-training for agents, including RL for open-ended tasks, goal-driven training for long-horizon trajectories, tool use, and efficient RL training and inference.
  • Enterprise environments and synthetic organizations, including scalable construction of realistic RL environments, simulation of multi-person collaboration, and automatic generation of tasks, skills, and rubrics.
  • Evaluation and rewards, including rubric-based grading, reliability of Large Language Models (LLM) judges, and evaluations that predict real-world value.
  • Learning from experts and from use, including learning from subject-matter-expert feedback and continual improvement after deployment.
  • Skills and organizational knowledge, including search, composing and tuning agent skills, and retrieval and embeddings.
  • Multimodal tuning, including adapting models to slides, documents, technical drawings, and an organization's style and voice.

This role suits PhD students who want research that is both novel and grounded in real needs, and who want to see their ideas move into products used by organizations worldwide. Interns will collaborate with researchers, applied scientists, engineers, and product managers, and are encouraged to publish their work.

Responsibilities

Research Interns apply inquiry and theory in practice, working alongside doctoral candidates and leading researchers. During the 12-week internship, interns are paired with mentors, collaborate with other interns and researchers, present findings, and take part in the research community. Internships are offered year-round, though they typically begin in the summer.

  • Design, implement, and evaluate new methods in training, environment, and evaluation pipelines, together with researchers, applied scientists, and engineers.
  • Turn real enterprise scenarios into well-posed research questions, and compare results against strong baselines, including the latest frontier models.
  • Document methods and findings, present results, demonstrate working prototypes, and prepare work for publication where appropriate.
  • Apply your knowledge and problem-solving skills to new challenges, with the chance to see your ideas realized in products used worldwide.
  • Embody Microsoft's culture and values.

Qualifications

Required: Currently enrolled in a Doctorate Degree program in Computer Science, Electrical Engineering, or a related STEM field.

Other requirements: Interns must be physically located in their manager's Microsoft worksite location for the duration of the internship. Applicants must submit a minimum of two reference letters, a cover letter, and any relevant work or research samples. Reference letters may be requested after the application is submitted, and may not be requested for all candidates. Applicants may wish to alert their letter writers in advance.

Preferred:

  • Research experience in reinforcement learning, post-training of large language models, LLM agents and tool use, evaluation, natural language processing, information retrieval, or multimodal models, shown through publications or open-source work.
  • Hands-on experience training, fine-tuning, or post-training large language or multimodal models, such as supervised fine-tuning, RL from human or AI feedback, preference optimization, or reinforcement fine-tuning.
  • Experience building environments, simulators, benchmarks, or evaluation methods for AI agents.
  • Programming skills in Python, experience with deep learning frameworks such as PyTorch, and experience running large-scale experiments.
  • Ability to turn open-ended, real-world problems into well-posed research questions and rigorous experiments.
  • Ability to think unconventionally to derive creative and innovative solutions.
  • Ability to collaborate effectively with your mentor and others on new research projects, with strong communication and cross-group collaboration skills.

Microsoft accepts applications and processes offers for these roles on an ongoing basis throughout the academic calendar (September to April). The position will remain open for a minimum of 5 days, with applications accepted until the position is filled.

Microsoft is an equal opportunity employer. All qualified applicants will receive consideration for employment without regard to protected characteristics. For accommodation requests during the application process, see the company's accessibility page.

Skills mentioned

  • Python
  • PyTorch
  • Deep Learning
  • NLP
  • Large Language Models
  • Reinforcement Learning

Related news: Microsoft expects 3,000 or fewer to accept buyout offer

More jobs at MicrosoftSee all 30

Research Intern in Redmond, WA

Microsoft · Redmond, WA

  • Internship
  • Hybrid

$82k – $162k /yr

4 days ago

Research Intern

Microsoft · Redmond, WA

  • Internship
  • Hybrid

$82k – $162k /yr

4 days ago

Browse more jobs

$82k – $162k /yr

Microsoft

Apply (opens in a new tab)