广告架构高级工程师-Ads ML Infra
Core
Design and support large-scale distributed training and inference for 100B+ parameter models with 1M sequence lengths, optimizing ML efficiency and supporting novel paradigms.
Role type
Senior IC Ads ML Infrastructure Engineer
Builds
High-performance distributed training and inference systems for advertising models
Domain
Advertising technology + Distributed Systems + Deep Learning
Deliverable
production ML models
Required skills
C/C++ programming, Linux environment expertise, distributed training frameworks (DDP/F SDP/TP/SP/PP), inference frameworks (RPC/VLLM/TRT), model optimization (quantization, pruning, compression, distillation, NAS, compilation), GPU performance analysis, large-scale distributed system architecture
Preferred skills
Reinforcement learning, Federated learning, Graph learning, Optimizers, Deep learning framework internals
Responsibilities
Implement and maintain distributed training and inference pipelines for massive models; Collaborate with algorithm teams on joint optimization; Optimize ML efficiency through model compression and system tuning; Support specialized ML paradigms like RL and FL
Seniority
Senior, hands-on IC