Member of Technical Staff, Systems Infrastructure

Responsibilities

• Design, build, and operate core infrastructure systems for large-scale training and inference.
• Build benchmarks, traces, and simulators to evaluate scheduling, caching, and network performance.
• Identify and eliminate bottlenecks across the stack, from kernels and drivers to scheduler policies.
• Turn systems research ideas into production systems that withstand real workloads, failures, and customers.
• Collaborate with research and inference teams to align infrastructure and model design.
• Track emerging hardware, interconnects, and systems research to inform technical direction.

Areas of Focus

• Scheduling and resource management: GPU job scheduling across heterogeneous hardware, multi-tenant isolation, fair sharing, preemption, topology- and locality-aware placement, autoscaling, fleet utilization, and capacity planning.
• Distributed storage and caching: Storage and caching for model weights, checkpoints, datasets, and KV cache; tiering across memory, local NVMe, and object storage; cache admission and eviction, consistency, cold-start, and weight-loading performance.
• Datacenter networking: RDMA/RoCE and InfiniBand, collective communication performance, congestion control, topology design, load balancing, tail latency, and reliability at fleet scale.

Requirements

• PhD completed within the last six months or expected by December 2026 in Computer Science, Computer Engineering, Electrical Engineering, or a related field.
• Research background in distributed systems, operating systems, scheduling and resource management, storage systems, computer networks, computer architecture, or high-performance computing.
• Depth in at least one focus area, demonstrated through a dissertation, publications, or systems built.
• Strong systems programming skills in C/C++, Rust, Go, or a similar language, plus working proficiency in Python.
• Experience building and evaluating real systems, not only simulations, and rigorously measuring performance.
• Ability to communicate systems designs and results clearly to varied audiences.

Preferred Qualifications

• First-authored publications at top-tier systems or networking venues.
• Infrastructure, cloud, or HPC internships, or meaningful open-source systems contributions.
• Experience with GPU clusters, CUDA, ROCm, Triton, or heterogeneous compute environments.
• Kubernetes and cloud infrastructure experience, or experience operating multi-tenant production clusters.
• Performance profiling and debugging experience in distributed environments.
• Research or engineering achievements demonstrated through grants, fellowships, patents, or systems competitions.