大模型训练调优工程师
Tasks
- Analyze training stability, convergence, and resource utilization
- Apply mixed precision and compilation optimizations for training performance
- Build and maintain multimodal model training pipeline
- Collaborate with algorithms on model architecture and training strategy
- Deploy multi GPU distributed training with DDP
- Design training quality evaluation and anomaly monitoring
- Implement parameter efficient fine tuning with LoRA and PEFT
- Optimize training pipeline with task scheduling and data loading
Perks/Benefits
- N/A
Skills/Tech-stack
Adapter | DDP | Data-parallel | DeepSpeed | Distributed Data Parallel | Distributed data | FSDP | Heterogeneous Acceleration | LoRA | Mixed Precision | Model Deployment | Multi-GPU | PEFT | PyTorch | Transformer
Education
Bachelor of Engineering | Bachelor of Science | Master of Science
Regions
Countries
States
Related jobs
-
Mid-level Full Time北京 R6h ago
-
Entry-level Full Time北京 R1d ago
-
Mid-level Full Time北京 R13d ago
-
Mid-level Full Time北京 R13d ago
-
AWS | Azure | JavaScript | NoSQL | Node.jsFast-paced environment | Remote workMid-level Full TimeHangzhou R1mo ago
-
AWS | Agile | Azure | Blockchain | CursorRemote workMid-level Full TimeShenzhen R1mo ago