编排调度高级研发工程师（ML Infra方向）-基础技术
Core
Design and build the underlying infrastructure for ByteDance's machine learning platform, focusing on orchestrating and scheduling ML workloads across offline unified schedulers.
Role type
Senior IC ML Infrastructure Engineer
Builds
Next-generation resource scheduling systems for large-scale Kubernetes/Yarn hybrid scenarios
Domain
Cloud-native infrastructure, distributed systems, GPU acceleration
Deliverable
infrastructure
Required skills
Kubernetes architecture, cloud-native system development, data structures and algorithms, GPU training acceleration (RDMA, Cuda, NCCL), GPU isolation technologies (EGPU, MIG), resource utilization optimization
Preferred skills
TensorFlow/PyTorch/MXNet frameworks, large-scale distributed system management, industry computing infrastructure trends
Responsibilities
Build and manage the bottom-layer infrastructure of the ML platform; Optimize offline task training resource flow efficiency; Design and construct resource scheduling systems for large-scale K8s/Yarn fusion scenarios