模型系统工程师--抖音推荐方向
Core
Design and develop high-performance, low-latency recommendation model engineering architectures for Douyin's recall and ranking scenarios, optimizing inference and training under massive throughput.
Role type
Senior IC model systems engineer (recommendation)
Builds
Unified model engineering solutions for recall and ranking, optimized inference/training pipelines, and end-to-end funnel efficiency improvements.
Domain
Internet / Recommendation Systems / High-Performance Computing
Deliverable
production ML models
Required skills
C/C++ or Python, data structures and algorithms, CUDA programming, distributed system design, machine learning frameworks (TensorFlow/PyTorch), GPU acceleration, multi-machine multi-card optimization, heterogeneous accelerator integration.
Preferred skills
ACM/ICPC/NOI/IOI competition awards, experience with MPS/MIP mixed deployment, hierarchical storage optimization.
Responsibilities
Provide efficient and stable model engineering solutions for Douyin's recall and ranking; optimize online inference and offline training performance via hardware-software co-design; explore heterogeneous acceleration and multi-node optimization based on Model as a Service; break through compute limits to improve end-to-end funnel efficiency from sample to inference.