Infrastructure Engineer (InfiniBand / NCCL) - Remote

Overview

We are seeking an Infrastructure Engineer with a focus on InfiniBand/NCCL to join our Infrastructure Engineering team. Our engineers design and build automation, tooling, and systems that bridge the gap between physical infrastructure and the platforms that power large-scale AI/ML and HPC workloads.

This role combines the breadth of a core infrastructure engineer with a specialty in high-performance networking and GPU communication. You’ll help ensure our InfiniBand fabric and NCCL stack are tuned, reliable, and efficient at scale — supporting some of the world’s largest GPU clusters.

This is a fully remote position, although candidates must be based in the continental United States. Unfortunately, we are unable to provide sponsorship for this role.

In Short

• Design, build, and maintain automation, APIs, and frameworks to manage physical infrastructure at scale.
• Develop and extend systems for server lifecycle management.
• Implement and tune InfiniBand networking and NCCL configurations for multi-GPU communication.
• Collaborate with Network, Platform, and Infrastructure Operations teams to support new infrastructure rollouts.
• Diagnose and improve performance across GPU, NVSwitch, PCIe, and InfiniBand layers.
• Write clear design documents and technical documentation to capture best practices.

Requirements

• 8+ years of professional experience in infrastructure engineering, HPC, or related domains.
• Strong experience with Linux in production environments.
• Proficiency in Python or similar languages for automation.
• Deep understanding of InfiniBand networking (CX7 HCAs, fabrics, partitioning, GPUDirect).
• Familiarity with NCCL, CUDA, and GPU topology optimization.
• Knowledge of containerization and orchestration concepts.
• Strong written and verbal communication skills.

Benefits

• Enjoy collaborating with a motivated, execution-focused team.
• Comfortable operating with autonomy while aligning to company objectives.
• Value precision, documentation, and knowledge-sharing.
• Excited to grow as both a domain specialist (InfiniBand/NCCL) and a generalist infrastructure engineer.