Network Engineer, InfiniBand and RoCE
Taipei, on-site
About Zettabyte
- AI computing company: we turn GPU hardware into reliable, high-utilisation AI compute through software, orchestration and full lifecycle operations
- What we offer: enterprise GPU cloud and AI inference, software for data centre owners to run their own GPU clouds, turnkey and sovereign AI data centres, edge computing and GPU financing
- Who we work with: GPU owners, data centre operators, enterprises, OEMs, integrators and financiers across multiple markets
- How we work: fast-moving and high-accountability; we value speed, ambition, clarity, ownership and measurable outcomes
Role Summary
The Network Engineer, InfiniBand and RoCE designs, builds and operates the high-performance fabrics that connect GPU clusters in Zettabyte's data centres and customer deployments.
The role covers the full fabric lifecycle, from topology design and cabling plans through bring-up, performance validation and day-to-day operations.
The Network Engineer works with data centre, hardware and platform teams so that every cluster delivers the bandwidth and latency that AI workloads require.
Reporting Line
Reports to: Head of AI Infrastructure
Key Responsibilities
- Fabric design: design InfiniBand and RoCE fabrics for GPU clusters, including topology, oversubscription, addressing and cabling plans.
- Bring-up: lead fabric installation and bring-up with data centre teams and vendors, including switch and adapter firmware.
- Performance validation: run bandwidth, latency and collective communication tests before clusters are handed to customers.
- Ethernet and front-end networks: support storage, management and front-end networks that sit alongside the GPU fabric.
- Operations: monitor fabric health, troubleshoot link and congestion issues, and lead network incident response.
- Configuration management: keep switch configuration, firmware and inventory under version control and automation.
- Vendor management: work with network vendors and OEM partners on hardware selection, defects and support cases.
- Standards: write reference designs and runbooks that make each new deployment faster than the last.
Required Qualifications
- 5+ years in data centre or HPC networking
- Hands-on experience with InfiniBand (NVIDIA Quantum) and RoCE fabrics, including UFM or equivalent management tools
- Strong understanding of RDMA, congestion control (PFC, ECN) and lossless Ethernet
- Experience with leaf-spine and fat-tree topologies at scale
- Scripting in Python or Bash for automation and testing
- Willingness to work on-site in data centres, including occasional work outside office hours
- Clear written English; working Mandarin is strongly preferred
Bonus Qualifications
- NVIDIA networking certifications or equivalent
- Experience with NCCL performance testing on multi-node GPU clusters
- Familiarity with Spectrum-X or other AI Ethernet platforms
- Experience with network automation tools such as Ansible or NetBox
What Success Looks Like (First 90 to 180 Days)
- A reference fabric design is documented and approved for new GPU clusters
- New clusters pass a standard fabric validation suite before handover
- Switch configuration and firmware are tracked in version control
- Fabric incidents have runbooks and clear escalation paths with vendors
Why This Role Matters
- Cluster performance: large-scale training and inference are only as fast as the fabric that connects the GPUs.
- Repeatability: strong network standards let Zettabyte deliver new clusters quickly and consistently.
Working Norms at Zettabyte
- Bias to action: deliver drafts quickly, then iterate on feedback
- Be precise: double-check numbers and facts; sloppy analysis creates downstream cost
- Make it usable: package outputs for decision-making, not data dumps
- Own the process: whoever drives a deliverable manages stakeholders and timelines
- Learn the ecosystem: understand GPUs, cluster operations and the buyer landscape
Details
- Workplace: On-site, 5 days a week, Taipei office
- Travel: regular visits to data centre sites in Taiwan, with occasional travel to other deployment sites
- Employment type: Full-time
- Compensation: Negotiable; regular monthly pay of NT$40,000 or more
- Benefits: Labour and health insurance; 6% labour pension contribution; generous paid leave with no unpaid period for new joiners; free snacks, drinks and coffee; employee share options after a qualifying period
- Right to work: Required in Taiwan
- Equal opportunity: We hire on merit, performance and role fit, and do not discriminate on any ground protected by Taiwan's Employment Service Act or Gender Equality in Employment Act (including age, gender, sexual orientation, marital status, appearance and origin)