Tasks:
- * Build shared storage using idle NVMe drives
- * Deploy storage solutions in production
- * Develop solutions for slow checkpointing
- * Evaluate and customize high-performance storage systems
- * Implement GPUDirect Storage
- * Improve distributed storage stability and SLA
- * Optimize FUSE clients, metadata services, and storage engines
- * Optimize small-file and dataset loading
- * Optimize storage over InfiniBand and RoCE
- * Optimize VFS and FUSE kernel interactions
- * Tune memory management
- * Tune object-storage caching
- * Tune RDMA protocol stacks
Perks/Benefits:
Skills/Tech stack required:
[AI Training] [AI training and inference] [Alluxio] [C#] [C++] [Ceph] [CUDA] [DAOS] [Distributed Storage] [DMA] [File systems] [Fuse] [Go] [GPUDirect Storage] [Infiniband] [JuiceFS] [Kernel bypass] [Linux Kernel] [Lustre] [NVMe] [Object storage] [PCIe P2P] [POSIX file systems] [Python] [RDMA] [RoCE] [Rust] [S3-compatible] [S3-compatible object storage] [SSD] [Training and inference] [VFS] [Zero copy]
Educational requirements:
N/A
Role(s):
[Distributed Storage Engineer] [Engineer] [HPC Storage Engineer] [Storage Engineer] [Storage Systems Engineer] [Systems Engineer]