PhD AI RL Infrastructure Intern: Post-Training

AMD; Santa Clara (CA); Hybrid; USD 55,000 - 69,000; Full time

Job summary

AMD invites PhD students to join a Summer 2027 internship in Santa Clara, CA, to develop RL-based post-training infrastructure for large language and multimodal models. You will work on scalable training pipelines, multi-GPU systems, and production-quality tooling, with researchers and industry leaders.

You will gain real-world experience, collaborate across teams, and contribute to reproducible research and publications, in a hybrid onsite setting at AMD's Santa Clara campus.

Qualifications

• Pursuing a PhD in a related field with strong programming skills.
• Experience with RL, LLM post-training, RLHF/RLAIF, or preference optimization.
• Familiarity with distributed training, multi-GPU workloads, and large-scale inference.

Responsibilities

• Develop and optimize infrastructure for RL-based post-training of large language and multimodal models.
• Build scalable systems for rollout generation, inference, reward computation, and policy updates.
• Improve distributed training efficiency, fault tolerance, and resource utilization.
• Design interfaces for researchers to implement RL algorithms quickly.
• Build tools for experiment configuration, logging, monitoring, and reproducibility.
• Profile training pipelines and resolve performance and memory bottlenecks.
• Support on-policy and off-policy workflows with verified or model-generated feedback.
• Collaborate with researchers to translate requirements into production-quality infrastructure.
• Document system designs and contribute to technical reports and publications.

Skills

Python

PyTorch

Reinforcement learning

Distributed training

Multi-GPU

Education

PhD in Computer Science or ML/AI/CE

Tools

Docker

Kubernetes

Job description

AMD invites PhD students to join a Summer 2027 internship in Santa Clara, CA, to develop RL-based post-training infrastructure for large language and multimodal models. You will work on scalable training pipelines, multi-GPU systems, and production-quality tooling, with researchers and industry leaders.

You will gain real-world experience, collaborate across teams, and contribute to reproducible research and publications, in a hybrid onsite setting at AMD's Santa Clara campus.