Software Engineer in Machine Learning Infra - TikTok Recommendation Architecture
Core
Building and optimizing the architecture for TikTok's recommendation system, focusing on system stability, high availability, and performance of online services and offline data flows.
Role type
Senior IC machine learning infrastructure engineer
Builds
Globalized large-scale recommendation system, high-performance online services, and scalable storage/computing systems
Domain
Internet / Recommendation Systems / Distributed Systems
Deliverable
infrastructure
Required skills
C/C++/golang, GPU hardware architecture, CUDA, cuDNN, GPU performance analysis, Linux environment
Preferred skills
Deep model inference/training, model optimization (TVM, MLIR, XLA), TensorFlow, PyTorch, MxNet
Responsibilities
Serve and train ML models, build high-performance online services, design globalized recommendation systems, integrate hardware and software systems, develop software solutions, update and enhance existing software, conduct software testing and validation
Seniority
Mid-level, hands-on IC