Director, Software Engineering

Director, Software Engineering Compute will lead the engineering organization that builds and operates our compute platform. You will evolve it into a scalable, efficient, and AI-driven compute model spanning private and public cloud, covering scheduling, capacity management, and performance engineering.

What you get to do in this role:

• Lead the compute engineering organization, providing technical leadership, mentoring, and career development for managers and engineers.
• Own the roadmap for the compute scheduler, including placement, bin-packing, prioritization, preemption, and fairness across workloads.
• Define and drive the compute-at-scale model, including unit economics such as cost per workload, utilization, and capacity headroom.
• Lead capacity management for private and public cloud: demand forecasting, reservations, procurement, and cross-cloud placement.
• Establish and scale a performance engineering practice: benchmarking, load and scale testing, regression detection, and SLOs for scheduling latency and throughput.
• Apply AI and ML to demand forecasting, anomaly detection, and placement optimization. Set the AI/ML roadmap and decide where it outperforms heuristics.
• Work with product management, SRE, FinOps, and infrastructure partners to set priorities and tie capacity decisions to cost and customer outcomes.
• Drive engineering excellence through architecture reviews, design standards, and modern software and DevOps practices.
• Anticipate technical and organizational risks that affect multiple products or platforms, and provide clear, structured mitigations.
• Assess availability, scalability, and operational risk across compute systems, and lead the optimization work that follows.
• Manage resources, staffing, and succession planning to build and retain a best-in-class engineering organization.
• Share best practices and engineering standards with other teams across the organization.
• Define success measures for the first 12 to 18 months, such as utilization gains, scheduler latency targets, forecast accuracy, and cost per unit.