below marketmost similar roles pay hereabove market
This listing doesn't post a salary. Most similar roles pay
$154,457–$246,125.
Based on 240 similar postings.
Employer
About Microsoft
Microsoft Corporation is a global technology leader producing software, hardware, and cloud services including Windows, Office 365, Azure cloud platform, Xbox gaming, and Surface devices. Industry: Software & Cloud Computing
Microsoft currently has
651 open roles
on FindRole.
Listed pay typically runs
$119,800–$234,700
across 605 roles with salary data.
Research Intern - Coding Agents & Post-Training Optimization
Research Interns join a team of doctoral candidates and researchers to advance research and development strides over a twelve-week internship. In this role, you will collaborate with mentors to present findings and contribute to the research community. Your daily work involves training or fine-tuning transformer models, specifically focusing on building code-generation models or coding agents. You will utilize Python and PyTorch while working with frameworks like TRL, verl, OpenRLHF, or DeepSpeed-Chat. The work addresses technical problems in LLM post-training, including RL, SFT, and preference optimization. You will also handle model compression, efficient inference, and quantization using tools like vLLM, SGLang, and TensorRT-LLM. Key tasks involve distributed training with FSDP or Megatron and GPU profiling using CUDA or Triton.
Contribute to research and development strides in computer science or related STEM fields.
Present research findings to mentors and other researchers.
Train or fine-tune transformer models using Python and PyTorch.
Implement RL for LLMs at scale using frameworks like TRL, verl, or DeepSpeed-Chat.
Build and train code-generation models or coding agents using standard benchmarks.
Perform model compression techniques including quantization, distillation, pruning, or speculative decoding.
Execute distributed training using FSDP, DeepSpeed, or Megatron frameworks.
Conduct GPU profiling or kernel work using CUDA and Triton.
What we're looking for
Currently enrolled full-time in a Doctoral Degree program in Computer Science, Computer Engineering, Electrical Engineering, or a related STEM field.
At least one additional quarter or semester of school remaining following the completion of the internship.
Research experience in LLM post-training, model compression, or efficient inference.
Proficient Python and PyTorch skills with hands-on experience training or fine-tuning transformer models.
First-author publications at top venues (NeurIPS, ICML, ICLR, ACL, EMNLP, MLSys, ICSE, FSE) or equivalent research impact.
Experience with RL for LLMs at scale using frameworks such as TRL, verl, OpenRLHF, or DeepSpeed-Chat.
Experience building or training code-generation models or coding agents with benchmarks such as SWE-bench, LiveCodeBench, Terminal-Bench, or HumanEval.
Hands-on work with quantization, distillation, pruning, or speculative decoding; familiarity with serving stacks such as vLLM, SGLang, TensorRT-LLM, or ONNX Runtime.
Large Language Models
Python
Deep Learning
Reinforcement Learning
Model Compression
Distillation
Dataset Curation
Model Alignment
Data Generation
Formal Methods
Verification
Python
Large Language Models
retrieval-augmented-generation
Graph Machine Learning
Computer Vision
Statistics
Matrix Analysis
Probability
Machine Learning
Graph Theory
Computer Science
Machine Learning
Natural Language Processing
Reinforcement Learning
Foundation Models
Artificial Intelligence
Human-AI Interaction
Computational Social Science
User Modeling and Simulation
Continual Learning
Python
PyTorch
Machine Learning
Natural Language Processing
Computer Vision
Multimodal AI
Large Language Models
Vision-Language Models
Document Intelligence
Deep Learning
AI Systems
Prototypes