N
Performance Engineering Manager
Не указаноПолная занятостьEngineering ManagementComputer HardwareSemiconductors
Lead performance strategy and optimization for large-scale AI GPU clusters and data center solutions. The role spans distributed training and inference, high-performance networking, DPU and storage use cases, and cluster observability.
Обязанности
- Set performance strategy, characterization, test plans, and optimization for AI GPU clusters running distributed training and inference workloads.
- Evaluate and optimize networking performance, including RDMA, networking protocols, collective communication, congestion control, and load balancing.
- Research and analyze DPU and storage performance for North-South use cases supporting AI inference.
- Develop performance observability strategy and dashboards using telemetry and real-time metrics across NICs, switches, GPUs, and NVLink.
- Lead root-cause analysis of multi-node performance bottlenecks and coordinate mitigation across hardware, firmware, and software teams.
Требования
- Bachelor’s or master’s degree in computer science, computer engineering, software engineering, or equivalent technical experience.
- At least 8 years of overall experience with deep expertise in high-performance networking, RDMA, and systems performance.
- At least 3 years managing engineering teams focused on technical performance or R&D.
- Hands-on experience analyzing and optimizing collective communication, such as NCCL or MPI, and network traffic for large-scale distributed AI workloads.
- Hands-on experience designing, deploying, and customizing Grafana dashboards for cluster monitoring, alerting, and data visualization.
- Strong analytical, communication, and cross-team leadership skills.
Будет плюсом
- Experience optimizing NCCL, RDMA/RoCEv2, or custom collective algorithms for large GPU deployments, including LLM and MoE workloads.
- Experience tuning adaptive routing, PFC/ECN congestion control, or packet spraying.
- Experience building automated performance tools, root-cause analysis agents, or regression frameworks.
- Experience developing Grafana plugins, advanced dashboard panels, or alert workflows using PromQL or LogQL.