Infrastructure Engineer Intern (TikTok Recommendation Architecture) - 2027 Start (PhD)
Core
Design and develop the Recommendation and Search system architecture for TikTok, ensuring stability, high availability, and performance optimization for online services and offline data streams.
Role type
PhD intern, infrastructure engineer (distributed systems & deep learning)
Builds
Recommendation middle-office, data middle-office, and high-performance GPU kernel architectures
Domain
Internet / Recommendation Systems / Deep Learning Infrastructure
Deliverable
production ML models
Required skills
C++, CUDA, Python, PyTorch, TensorFlow, distributed training, GPU kernel optimization, parallel computing, computer architecture
Preferred skills
DeepSpeed, FSDP, Megatron-LM, vLLM, TensorRT-LLM, SGLang, LLM inference/serving systems
Responsibilities
Conduct novel research on distributed training and inference system optimization for large-scale recommendation models and LLMs; Design and implement high-performance GPU kernel architectures and communication primitives to accelerate deep learning workloads; Publish original research findings at top-tier academic conferences
Seniority
Intern, research & development