Research Intern - AI /Machine Learning Numerics & Efficiency

Microsoft•Redmond, WA
•$6,810 - $14,610•Onsite

About The Position

As part of Microsoft's Systems Planning and Architecture (SPARC) group, the Research Intern will contribute to research and exploration in advanced machine learning systems, focusing on the numerics, data types, and compute technologies behind next-generation AI workloads at Azure scale. The intern will collaborate across Azure teams on model-efficiency approaches, including low-precision formats, quantization strategies, machine-learning kernel development, and benchmarking and analysis of emerging model architectures and hardware capabilities. The role involves evaluating, prototyping, and analyzing algorithmic and numerical techniques to improve the performance, cost, and efficiency of training and inference for large-scale models. The work can inform decisions about compute platforms, acceleration strategies, and system-level optimizations, including internal silicon efforts, software runtimes, and industry partnerships. The internship offers experience in ML systems, emerging data types, kernel optimization, reinforcement learning, performance modeling, and Azure AI and hardware technologies. This is a 12-week research internship. Interns are paired with mentors, collaborate with other interns and researchers, present findings, and contribute to the research community. Interns must be physically located at their manager's Microsoft worksite for the duration of the internship. Research internships are offered year-round and typically begin in summer.

Requirements

  • Currently enrolled in a doctoral degree program in Computer Science, Electrical Engineering, or a related STEM field.
  • Must be physically located at the manager's Microsoft worksite for the duration of the internship.
  • Submit at least two reference letters, a cover letter, and relevant work or research samples.

Nice To Haves

  • Experience programming in Python and/or C/C++.
  • Strong understanding of transformer-based model architectures, including attention mechanisms, KV cache behavior, and common training and inference bottlenecks.
  • Experience with modern machine-learning frameworks and runtimes such as PyTorch, Hugging Face Transformers, SGLang, or vLLM.
  • Expertise in GPU/NPU kernel development using CUDA, Triton, ROCm, or comparable frameworks, and fast model bring-up on a new stack.
  • Working knowledge of low-precision numerics (FP8, INT8/4, MX formats, NVFP4, etc.), quantization methods, or hardware-software co-design for large-scale model efficiency.
  • Practical experience evaluating large-scale models, including benchmarking state-of-the-art LLMs and fine-tuning large models using SFT or RL.
  • Analytical and problem-solving skills, with an interest in machine-learning systems and computational performance.
  • Strong communication and collaboration skills, including the ability to work effectively across research and engineering teams.

Responsibilities

  • Research and explore advanced machine-learning systems, numerics, data types, and compute technologies for AI workloads at Azure scale.
  • Investigate model-efficiency approaches, including low-precision formats, quantization strategies, and machine-learning kernel development.
  • Benchmark and analyze emerging model architectures and hardware capabilities.
  • Evaluate, prototype, and analyze algorithmic and numerical techniques to improve the performance, cost, and efficiency of large-scale model training and inference.
  • Collaborate with Azure teams, mentors, other research interns, researchers, and engineering teams.
  • Present research findings and contribute to the research community.
© 2026 Teal Labs, Inc
Privacy PolicyTerms of Service