Tech - HPC and AI Storage Engineer

 Direct experience with IBM ESS/SSS: at least three years of specific practical experience in the administration of IBM Elastic Storage System (ESS) or Storage Scale System (SSS) devices. Mastery of executing specialized platform commands, such as mmvdisk, mmhealth, and mmchrecoverygroup.

 Advanced knowledge of high-performance S3: demonstrable experience in the design of enterprise S3 object storage nodes. Expert knowledge of CES S3 bucket structures, policy application, multi-node load balancing, and performance analysis using tools such as COSBench.

 Mastery of parallel file systems: deep knowledge of the internal architecture of IBM Storage Scale -formerly GPFS-, including token management, pagepool definition, and distributed locking mechanisms.

 Experience in all-flash infrastructures: solid knowledge of NVMe/PCIe Gen 5 storage paths, kernel block layer optimization, and direct I/O mechanisms.

 Advanced knowledge of high-performance networking: expert-level experience in the administration of RDMA networks, Subnet Managers, adaptive routing, and low-latency storage network design.

• Advanced English level.

To perform the position, the skills that would fit with the team and the project would be:

• Ability to work as part of a team.

Telefónica Tech is the leading digital transformation company of the Telefónica Group. We offer a wide range of integrated technological services and solutions in Cybersecurity, Cloud, IoT, Big Data, Artificial Intelligence, and Blockchain, with which we support our clients in their digital transformation.

We are a group of over 6,200 brave people who work every day from different points around the world to achieve excellence, through leadership based on transparency and team spirit. If you identify with our pillars, we can't wait to meet you!

www.telefonicatech.com

We are looking for an expert storage professional for HPC and artificial intelligence to act as the primary person responsible for the architecture and administration of our IBM Storage Scale System 6000 (ESS 6000) all-flash infrastructure.

In this position, you will be responsible for maximizing the multi-TB/s transfer capacity of our multi-petabyte ESS platform, providing the necessary performance for high data consumption artificial intelligence training clusters and complex high-performance computing (HPC) environments.

This position connects hardware-optimized parallel storage with container-based cloud workloads. You will lead the integration of high-performance S3 object storage via Cluster Export Services (CES), ensuring an efficient data flow toward high-density NVMe tiers and multi-node GPU clusters, making the most of the infrastructure's capacity.

• Administration of the ESS 6000 platform and hardware: deploy, expand, and maintain IBM ESS 6000 systems. Take responsibility for container-based software updates and for monitoring the hardware status in the system's two server modules.
• Storage Scale RAID and erasure coding management: design, format, and optimize Storage Scale RAID configurations. Manage recovery groups, virtual disks -vdisks-, and physical disks -pdisks-, using declustered erasure coding (declustered erasure coding) to ensure extreme write performance and reduce reconstruction times.
• Hierarchical Storage Management (HSM) and tape tiering: design, implement, and maintain transparent tape storage policies using IBM Storage Archive -formerly Spectrum Archive / Linear Tape File System (LTFS)-. Automate the transparent migration and retrieval of data between the ESS flash tier and the tape infrastructure.
• Enterprise tape library administration: manage high-end IBM libraries, such as TS4500, incorporating enterprise-level LTO or TS1100 drive technologies. Supervise library partitions, cartridge lifecycle, drive access path configurations, and SAN zoning via Fibre Channel or SAS.
• High-performance S3 and multi-protocol access: deploy, expand, and monitor NooBaa-based CES S3 clusters on specific protocol nodes. Optimize multi-protocol concurrency to allow the same underlying ESS NVMe blocks to be read via native S3 APIs, POSIX, or NFS, without data corruption or significant latency penalties.
• High-performance networking and direct transport engineering: administer 400 Gb NDR or 200 Gb HDR InfiniBand/RoCE networks via the ConnectX-7 adapters integrated into the platform. Deploy and validate NVIDIA GPUDirect Storage (GDS) and S3 over RDMA to transfer data directly from the ESS 6000 flash tier to GPU memory.
• Orchestration of HPC and AI data flows: align storage operations with parallel job schedulers, such as Slurm, and container storage interfaces (CSI) in Kubernetes or OpenShift. Use Active File Management (AFM) to provide low-latency caching and automate lifecycle management and tiering.

AND FOR THIS, WE BELIEVE IT WOULD BE IDEAL IF YOU HAD...