AI Infrastructure Engineer Intern (TikTok Live Recommendation Architecture) - 2027 Start
Core
Building scalable systems to support innovative algorithms for live broadcast recommendation, focusing on system stability, high availability, and performance optimization.
Role type
AI Infrastructure Engineer Intern
Builds
Scalable high-performance storage and computing systems for live streaming workloads
Domain
Live streaming / Recommendation systems
Deliverable
production ML models
Required skills
C++, CUDA, Python, GPU architecture, distributed training, deep learning frameworks, recommendation systems
Preferred skills
Production-grade training/inference systems for large-scale models, memory efficiency optimization, latency reduction, distributed training frameworks (NCCL, Horovod, DeepSpeed, FSDP)
Technologies
GPU, NCCL, Horovod, DeepSpeed, FSDP
Responsibilities
Optimize recommendation model memory usage and inference latency on GPU-based heterogeneous computing systems; Design and implement high-throughput distributed training and inference pipelines; Perform GPU kernel profiling and performance tuning for real-time inference.
Seniority
Intern
