Overview

• Assess AI training and inference workloads and translate business and technical requirements into
infrastructure architecture, system specifications and capacity plans.
• Evaluate heterogeneous computing platforms, including Intel and AMD CPUs, NVIDIA GPUs, and
accelerators such as Ascend NPUs and Google TPUs, to determine workload suitability and identify
potential performance bottlenecks.
• Develop plans for computing resource pooling, infrastructure expansion and long-term capacity
requirements, considering performance, scalability, compatibility and power consumption.
• Design and integrate GPU servers, high-performance storage, RDMA networks, distributed training
frameworks, and containerization and virtualization platforms into coordinated infrastructure
solutions.
• Research, design and evaluate network architectures and communication topologies to support
efficient data transfer and distributed computing across cluster nodes.
• Conduct system integration, compatibility and performance testing; document findings and
recommend improvements to infrastructure configurations.
• Optimize CPU and accelerator utilization, memory bandwidth, storage throughput and inter-node
communication to improve training throughput, reduce inference latency and increase overall
resource efficiency.
• Oversee the technical operation of IT infrastructure, monitor system availability and performance,
investigate bottlenecks and coordinate corrective actions.
• Lead and coordinate technical personnel and service providers in the implementation, integration and
optimization of infrastructure solutions.
• Provide advanced technical advice to clients and internal teams on infrastructure selection,
deployment, upgrades and performance improvement.
• Prepare and maintain architecture diagrams, configuration specifications, capacity assessments, test
reports and operational documentation.