HPC Network Engineer (AI)

TL;DR

HPC Network Engineer (AI): Designing and optimizing high-performance network fabrics for HPC and AI workloads with an accent on low-latency, RDMA, and congestion control. Focus on building scalable InfiniBand/RoCE architectures, automating fabric provisioning, and solving complex latency and microburst challenges.

Company

hirify.global provides specialized high-performance computing and network operational services for demanding AI and compute workloads.

What you will do

• Design and evolve InfiniBand and Ethernet/RoCE architectures, including spine-leaf and rail-optimized fabrics.
• Configure and manage switches, HCAs, and fabric services to optimize latency and throughput.
• Lead performance validation through benchmarking, packet analysis, and root-cause analysis of congestion.
• Implement automation for provisioning and configuration drift detection using Python and Ansible.
• Build observability stacks for fabric health monitoring tied to SLOs.
• Collaborate with platform and workload teams to align network design with GPU cluster and storage needs.

Requirements

• 3–7+ years of network engineering experience in HPC, low-latency trading, or large-scale compute environments.
• Strong expertise in InfiniBand, RDMA (verbs), and RoCE (v1/v2) implementation.
• Proficiency in Linux networking fundamentals, including routing, VLANs, and MTU tuning.
• Ability to automate operations using Python and/or Bash and Ansible.
• Experience with NVIDIA/Mellanox utilities and firmware lifecycle management.
• English: C1 (Fluent) required for technical documentation and cross-team coordination.

Nice to have

• Experience with GPU clusters and NCCL/RDMA path awareness.
• Knowledge of parallel storage networking (e.g., Lustre/GPFS).
• Experience with Prometheus, Grafana, or ELK for fabric telemetry.

Culture & Benefits

• Hands-on ownership of high-performance network fabric design and operational reliability.
• Data-driven approach to improvements using telemetry and packet captures.
• Collaborative environment working closely with HPC/cluster engineers and application teams.