GPU AIDC/Data Center Operations Engineer

2일 전

Seoul, 대한민국 ThunderSoft 정규직
We’re hiring: GPU Data Center Operations Engineer | South Korea | We’re looking for hands-on engineers to keep GPU clusters running reliably at an AI data center in South Korea. You’ll work on site with GPU servers, high-speed networks, and storage systems that support AI training and inference.

Location:
On site at a data center in South Korea Schedule: Rotating shifts Languages: Working proficiency in both Korean and English What you’ll do
• Monitor cluster health, carry out equipment inspections, and respond to incidents during your shift.
• Rack, cable, configure, and deploy GPU servers. Diagnose issues such as GPUs disappearing from the system, GPU memory errors, abnormal power draw, PCIe or NVLink faults, and unexpected server reboots. Replace faulty components and verify recovery.
• Maintain Ubuntu Linux systems and troubleshoot InfiniBand and RoCE networks, including link failures, port negotiation issues, and packet loss.
• Install and troubleshoot NVIDIA drivers, CUDA, NCCL, and RDMA software. Work with engineering teams to test cluster performance and investigate incidents affecting training or inference.
• Keep asset records, cabling documentation, and support tickets up to date. Coordinate warranty repairs and firmware upgrades with vendors, and document fixes for future shifts. What you’ll bring
• A bachelor’s degree or higher and 1–3 years of hands-on, on-site experience in a data center, GPU cluster, or high-performance computing (HPC) environment.
• Practical Ubuntu administration skills, including service management, basic kernel and boot troubleshooting (initramfs, GRUB), and network configuration with Netplan, NIC bonding, VLANs, and routing.
• Experience troubleshooting Mellanox network adapters and InfiniBand or RoCE connections using tools such as ibstat, ibdev2netdev, and ibportstate.
• The ability to isolate hardware faults and replace components in the field, including GPUs, CPUs, memory, drives, power supplies, and high-speed network adapters.
• The ability to coordinate incident response and handovers in Korean and English, and the willingness to work on site on a rotating shift schedule. Hands-on experience with NVIDIA A100 or H100 clusters, RDMA, or production HPC operations is a plus. If this sounds like your kind of work, please send me a LinkedIn message with your CV and a brief note about the GPU or HPC systems you’ve supported. Referrals are welcome.