About the Role:Crusoe's Managed AI team builds the services customers use to customize and run large language models without managing infrastructure. On the training side, it's a post-training platform where customers adapt models to their own data with supervised fine-tuning (SFT) and reinforcement learning (RL). On the inference side, it's managed serving that runs open-weight and fine-tuned models, including LoRA adapters, fast and cost-efficiently on Crusoe's GPU fleet. The two are closely connected: models trained on the platform are deployed and served on the same infrastructure, and RL training depends on high-throughput inference for rollouts.
As an AI Applied Research PhD Intern, you'll own a research question somewhere on that path from training to serving. It could be making RL work reliably on real customer tasks, or making inference faster and cheaper without losing quality. You'll run experiments on frontier open-weight models (Qwen, GLM, DeepSeek and Kimi-class models) on large GPU clusters, turn your findings into recipes and platform features that customers use, and present the results to the team and leadership. This is a 12-week, full-time, in-person internship. Candidates who want to see their research run in production, not just in a paper, will find it a good fit.
What You'll Bring to the Team:- Education: Currently enrolled in a PhD program in Computer Science, Machine Learning, Electrical Engineering or a related field, with an expected graduation date between Fall 2027 and Spring 2029.
- Research experience: 2+ years of research experience in machine learning or ML systems, including hands-on work with large language models: training or fine-tuning (SFT, RLHF, DPO, GRPO/PPO) or inference optimization.
- Programming: Strong Python and PyTorch; comfortable reading and modifying large ML codebases.
- Experimental rigor: You design controlled experiments, choose the right baselines and metrics, and can tell a real effect from noise.
- Systems awareness: Working knowledge of how LLM training and inference run on GPUs, such as batching, memory, mixed precision and parallelism.
- Communication: You explain complex results clearly in writing and in person, and you work well with engineers turning research into product.
Bonus Points- Publications at venues such as NeurIPS, ICML, ICLR, MLSys, ACL, EMNLP or COLM on RL for LLMs, reward modeling, agents, efficient training or efficient inference.
- Hands-on experience with inference engines (vLLM, SGLang, TensorRT-LLM) or training frameworks (verl, OpenRLHF, TRL, Megatron-LM, DeepSpeed).
- Experience with quantization, speculative decoding, GPU kernels (CUDA, Triton) or LoRA/PEFT methods.
- Experience building agent environments, tool-use sandboxes or evaluation harnesses.
- Open-source contributions to Training/Infernece libraries.
Our Internship ProgramOur 12-week, in-person program pairs high-impact work with one-on-one mentorship and direct access to leaders across the organization. You'll gain hands-on experience that goes beyond the classroom. The program is built to support your professional growth and help you forge valuable connections within Crusoe and beyond.
We're looking for motivated, talented people who are eager to take on real responsibility and make a real impact. Are you ready to join the Future Cru?
2027 Summer Internship Dates (choose one cohort):- May 24 - August 13, 2027
- June 14 - September 3, 2027
Benefits:- Compensation will be $80/hour
- One-time housing stipend
- Access to HealthiestYou and Calm
- Paid holiday and volunteer days
- Commuter allowance