MiMo-大模型训练框架开发工程师
Tasks
- Accelerate data loading and preprocessing
- Build performance monitoring system
- Debug and optimize overlap of communication and computation
- Design training framework
- Develop and optimize model training
- Develop with Python and C++
- Enhance model scalability
- Identify performance bottlenecks
- Improve training efficiency
- Improve training stability
- Monitor training metrics in real-time
- Optimize memory and GPU utilization
- Perform performance evaluation and tuning
- Resolve distributed training communication issues
- Tune parallelism strategies
Perks/Benefits
- N/A
Skills/Tech-stack
C++ | CI/CD | Data Preprocessing | Data loading | DeepSpeed | Distributed Training | GPU Memory Optimization | GPU memory | Infiniband | Megatron-LM | Memory Optimization | NCCL | Nsight | PyTorch | Python | RoCEv2
Education
N/A
Related jobs
-
Entry-level Full Time北京 R4h ago
-
Mid-level Full Time北京 R4h ago
-
具身智能算法工程师-模型 CNY 500K-500KActor-critic | Deep learning | Distributed Training | Implicit Q Learning | Inference accelerationMid-level Full Time北京 R4h ago
-
AI基础设施研发工程师（Sandbox / 容器化)-MiMo CNY 180K-420KAppArmor | Argo Workflows | CPU resource scheduling | Cgroup | ContainerdMid-level Full Time北京 R4h ago
-
Lead Embedded Software Engineer CNY 349K-437KARM | BLE | C# | C++ | Embedded LinuxHybrid work model | Remote-friendly | Work from homeSenior-level Full TimeSuzhou, China R4d ago
-
Mid-level Full Time上海、深圳 R10d ago
-
模型部署与推理优化工程师 CNY 180K-360KC++ | Edge inference | Inference Performance | Inference Performance Optimization | Model DistillationMid-level Full Time北京 R16d ago
-
Entry-level Internship上海 R17d ago
-
Mid-level Full Time北京 R1mo 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