强化学习训练框架工程师（火山引擎机器学习平台）-Data AML
Core
Design, develop, and optimize the veRL training framework and veOmini LLM/DiT training modules to support efficient post-training and reinforcement learning for hot models like Wan2.1, Qwen, and DeepSeek.
Role type
Senior IC machine-learning engineer (training framework)
Builds
veRL and veOmini training frameworks for LLM and DiT models
Domain
Large-scale distributed systems and large language model training
Deliverable
production ML models
Required skills
C/C++, Python, PyTorch, distributed system development, RPC/NCCL/MPI, model training performance tuning
Preferred skills
Reinforcement learning framework development, open-source community contributions
Technologies
veRL, veOmini, Megatron-LM, DeepSpeed, TorchTuner, vLLM, SG-Lang, TensorRT, Wan2.1, Qwen, DeepSeek
Sourced via bytedance · Listed on CareerPlan, which tracks 854,000+ jobs from 20+ sources.
