Director of AI Infrastructure

Responsibilities

• Oversee the availability and performance of dense on-prem GPU clusters. Partner with hardware vendors and internal teams to ensure physical infrastructure meets demands of frontier model training.
• Direct strategy for Beaker orchestration platform. Optimize job scheduling for high utilization of on-prem assets and elastic cloud resources (AWS/GCP).
• Develop and execute long-term roadmap for storage balancing high-throughput performance for active training with cost-effective durability for petascale research data.
• Act as primary steward of GPU compute budget. Make data-driven decisions on bursting to cloud vs. investing in on-prem capacity.
• Serve as technical bridge to research teams, ensuring infrastructure accelerates diverse research objectives.

Requirements

• 12+ years in infrastructure, systems engineering, or HPC, with at least 5 years in leadership managing multi-disciplinary engineering teams.
• Bachelor’s degree in related field; relevant advanced degree may substitute for equivalent experience.
• Direct experience managing large-scale NVIDIA GPU clusters and high-performance networking (InfiniBand/RoCE).
• Strong background in Kubernetes, Slurm, or similar orchestration frameworks in hybrid-cloud configurations.
• Experience with distributed filesystems (e.g., WEKA, Ceph, Lustre) and cloud storage integration at scale.
• Proficient in Go or Python, with ability to review architecture and code for internal tooling.

Compensation: Base salary $176,400 - $264,600 plus generous bonus plans.