大模型训练系统工程师-火山方舟
Core
Design and develop training systems for the Volcano Ark platform to support Serverless fine-tuning (SFT) and reinforcement learning (RL) for large models.
Role type
Senior IC large model training systems engineer
Builds
Elastic training solutions for multi-tenant scenarios across multi-datacenter and heterogeneous hardware; reinforcement learning systems and user-friendly training interfaces.
Domain
Cloud infrastructure + Large Language Model training
Deliverable
production ML models
Required skills
Python, Rust, C++, framework design, distributed systems development, performance optimization, scientific modeling
Preferred skills
Heterogeneous hardware architecture optimization, high-performance networking, reinforcement learning frameworks
Responsibilities
Design elastic training schemes for complex multi-tenant scenarios; optimize training throughput and stability across multi-datacenter environments; design reinforcement learning systems to improve efficiency.
