AI Infra Optimization Engineer Graduate (TikTok Global E-Commerce Recommendation & Search Architecture) - 2027 Start
Core
Build and optimize infrastructure for LLM-based recommendation systems, focusing on training/inference acceleration, model scale-up, and end-to-end generative recommendation architectures.
Role type
Graduate AI Infrastructure Engineer (LLM & Recommendation)
Builds
Scalable recommendation foundation models and next-generation system architectures for e-commerce live-streaming, short videos, and commodity recommendation.
Domain
E-commerce, Recommendation Systems, Large Language Models (LLM)
Deliverable
production ML models
Required skills
C++, CUDA, Triton, Python, GPU architecture knowledge, distributed training, LLM training/inference, model optimization, PyTorch
Preferred skills
TensorRT, CUTLASS, model quantization, graph compilation, KV cache optimization, prefill/decode disaggregation, Megatron-LM, DeepSpeed, vLLM, SGLang, TensorRT-LLM, reinforcement learning
Technologies
PyTorch, C++, CUDA, Triton, TensorRT, Megatron-LM, DeepSpeed, vLLM, SGLang, TensorRT-LLM
Responsibilities
Develop and optimize LLM training/inference techniques for recommendation foundation models; design next-gen model infra architectures using open-source components; co-design end-to-end generative recommendation technologies; optimize GPU kernels and multi-GPU parallelism; research LLM-native recommendation models and architectures.
Seniority
Graduate (Entry-level)