AI Infrastructure Engineer Graduate (TikTok Live Recommendation Architecture) - 2027 Start
Core
Building and optimizing high-performance GPU-based systems for live broadcast recommendation model training and inference.
Role type
Graduate AI Infrastructure Engineer
Builds
Scalable storage and computing systems for live streaming recommendation algorithms
Domain
Live streaming / Recommendation Systems / GPU Computing
Deliverable
production ML models
Required skills
C++, CUDA, Python, GPU architecture knowledge, distributed training, system performance profiling, high-throughput pipeline deployment
Preferred skills
Deep learning compiler frameworks (TVM, LLVM), NCCL, Horovod, DeepSpeed, FSDP, open-source contributions, real-time system experience
Technologies
CUDA, NCCL, Horovod, DeepSpeed, FSDP, TVM, LLVM, C++, Python
Responsibilities
Optimize model performance and memory efficiency on GPU-based systems; Deploy and scale high-throughput training and inference pipelines; Develop tools and compiler optimizations for deep learning workloads; Analyze system performance via GPU profiling and kernel analysis; Build and maintain high-performance computing frameworks for model training and inference
Seniority
Graduate (Entry-level)