Graduate Research Engineer, AI Training Systems

ByteDance; Seattle (WA); On-site; USD 242,000 - 456,000; Full time

Job summary

ByteDance is seeking a Research Engineer Graduate for AI Training Systems Reliability & Performance on Seed Infra in Seattle. This role targets graduates with a PhD and strong coding skills in C++ and Python, with deep experience in PyTorch and distributed training across multi-GPU/multi-node environments.

You will work on improving reliability and performance of large-scale training systems, build profiling tools, and collaborate with model and infrastructure teams to scale efficiency.

Qualifications

• PhD or near completion in CS/EE/physics or related field.
• Strong programming skills in C++ and Python.
• Solid understanding of PyTorch training workflows and distributed runtime behavior.
• Familiarity with CUDA execution, NCCL, and GPU systems fundamentals.

Responsibilities

• Improve reliability and performance of large-scale training systems across pre-training, fine-tuning, evaluation, and inference.
• Build observability, profiling, and debugging tools for distributed ML workloads.
• Identify and optimize performance bottlenecks across GPU, networking, and storage layers.
• Contribute to distributed training frameworks in multi-GPU and multi-node environments.
• Collaborate with model and infrastructure teams to improve system scalability and efficiency.
• Support incident analysis and operational stability.

Skills

C++

Python

PyTorch

CUDA/NCCL

Distributed training

Education

PhD in CS/EE/Physics

Tools

Nsight

torch.profiler

Job description

ByteDance is seeking a Research Engineer Graduate for AI Training Systems Reliability & Performance on Seed Infra in Seattle. This role targets graduates with a PhD and strong coding skills in C++ and Python, with deep experience in PyTorch and distributed training across multi-GPU/multi-node environments.

You will work on improving reliability and performance of large-scale training systems, build profiling tools, and collaborate with model and infrastructure teams to scale efficiency.