High-Performance Distributed Storage Systems Intern

Please note: This position requires candidates to be proficient in using Chinese as the working language.

 

Compensation: This is a paid position. Compensation will be provided in accordance with company policy. Details will be discussed with qualified candidates during the interview process.

 

This role focuses on storage and data movement for data-intensive and model-intensive workloads. Typical workloads include large datasets, model checkpoints, factor data, and highly concurrent read/write paths.

The relevant questions are system-level: how throughput, tail latency, CPU overhead, DMA paths, protocol overhead, queueing, data layout, and device behavior interact under real load.

You will

• Characterize access patterns for training data, research datasets, factor stores, and checkpoint traffic

• Trace critical paths from userspace through the filesystem, page cache, block layer, network stack, and storage devices

• Evaluate RDMA, NVMe-oF, SPDK, and GPUDirect under realistic workload and deployment constraints

• Study how caching, data layout, queueing structure, zero-copy paths, and protocol choices affect end-to-end performance

• Design and prototype interfaces and implementations for a high-throughput, scalable storage substrate

• Validate design choices using traces, benchmarks, and workload replay

 

You should have

• Strong modern C++

• Strong grounding in either the Linux storage subsystem or RDMA networking

• Systems programming experience and a disciplined approach to performance analysis

• Interest in storage, networking, DMA paths, or workload-driven system design