AI Infra存储架构师-成都/上海
Core
Design and implement hierarchical storage architectures for inference systems, managing KVCache placement and eviction across HBM, DRAM, SSD, and distributed persistent storage to balance hit rates, latency, and cost.
Role type
Senior IC AI Infrastructure Storage Architect
Builds
High-throughput, low-latency distributed persistent storage systems supporting KVCache offloading, model weight, and checkpoint I/O
Domain
AI Infrastructure / Distributed Systems / Storage
Deliverable
production ML models
Required skills
C++, Rust, Python, Distributed Systems, Consistency Protocols, Metadata Services, Fault Recovery, NVMe SSD, RDMA, io_uring, SPDK, GPUDirect Storage, Parallel Computing, Memory Management
Preferred skills
Large Model Inference Systems, vLLM, SGLang, 3FS, Mooncake, LMCache, Ceph, DAOS, Hardware Architecture, Open Source Contributions
Responsibilities
Design hierarchical storage architectures for inference systems; Develop distributed persistent storage systems for KVCache and model weights; Build high-availability metadata services; Optimize end-to-end I/O paths using RDMA and zero-copy techniques; Ensure data reliability and stability.
Seniority
Senior, hands-on IC
