Researcher Intern - CoreAI Post Training
 Microsoft | |
 United States, Washington, Redmond | |
 Oct 29, 2025 | |
|
OverviewCome build community, explore your passions and do your best work at Microsoft with thousands of university interns from every corner of the world. This opportunity will allow you to bring your aspirations, talent, potential - and excitement for the journey ahead. The Microsoft CoreAI Post-Training team advances post-training methods for both OpenAI and open-source models. We work on continual pre-training, large-scale deep RL on large GPU fleets, data curation/synthesis at scale, and practical fine-tuning for research and product. We also build language + multimodal technologies used across Microsoft, with a special focus on code-centric models for GitHub Copilot and Visual Studio Code (completion and SWE agent models). Our work connects to efforts such as LoRA, DeBERTa, Oscar, Rho-1, Florence, and the open-source Phi family. We prize research innovation and bold experimentation-aiming for breakthroughs that materially advance the state of the art and ship into products. As an intern at Microsoft, you're stepping into a world of real impact from day one. You'll collaborate with global teams on meaningful projects, explore cutting-edge technologies like AI, and kick start your career while doing it. With a strong focus on learning and development, this is your opportunity to grow your skills, build community, and shape your future-all while being supported every step of the way. Microsoft's mission is to empower every person and every organization on the planet to achieve more. As employees we come together with a growth mindset, innovate and empower others, and collaborate to realize our shared goals. Each day we build on our values of respect, integrity, and accountability to create a culture of inclusion where everyone can thrive at work and beyond.
ResponsibilitiesDesign & evaluate datasets: build high-quality datasets/benchmarks; run ablations to measure impact and improve data effectiveness Advance model training: contribute to pre-training, post-training, and RL for language and multimodal models Develop data infrastructure: extend pipelines for ingest, preprocess, filter, and annotate large, heterogeneous data Data quality & analysis: assess text, image, video, audio, and code data for quality, diversity, and relevance; propose improvements Tooling & workflows: create lightweight tools for dataset auditing, visualization, and versioning to speed iteration Research & collaboration: work with researchers/engineers to push research and product boundaries with measurable impact | |