AIDC (AI Data Center) Architect (pre-sales)

Role Overview

SenseTime Hong Kong is seeking an experienced AIDC Architect to lead the technical architecture and delivery of next-generation AI Data Center infrastructure across Hong Kong.

The role will focus on large-scale GPU clusters, high-performance networking, ultra-high-density power, liquid cooling, PUE optimization and AI compute performance. The Architect will work closely with SenseCore, facility operators, technology vendors and customers to build scalable infrastructure supporting foundation-model training, inference and enterprise AI workloads.

This is a strategic role supporting SenseTime's growing AIDC business. Current Hong Kong initiatives already contemplate large scale mixed GPU infrastructure, including projects in the ~20 MW / 1,000+ PFLOPS range.

Key Responsibilities

• Lead end-to-end AIDC architecture design, covering GPU compute, network, storage, power, cooling and AI platform integration.
• Design and optimize large-scale GPU/HPC clusters for AI training and inference workloads.
• Architect high-performance fabrics using InfiniBand and/or RoCE, including RDMA and 200G/400G/800G networking.
• Define high-density rack architecture, power capacity and Direct Liquid Cooling (DLC) requirements.
• Drive PUE, PFLOPS/MW, GPU utilization and overall compute-efficiency optimization.
• Conduct cluster benchmarking and identify performance bottlenecks across GPU, network, storage and software layers.
• Work with SenseCore teams on cluster orchestration, scheduling, monitoring and AI workload deployment.
• Coordinate with data-center operators, utilities, MEP consultants, GPU/server OEMs, networking and cooling vendors.
• Review technical designs, Bills of Materials, commissioning plans and infrastructure acceptance tests.
• Support customer technical workshops, solution design, AIDC proposals, RFPs and capacity planning.
• Develop reusable SenseTime AIDC reference architectures and technical standards for expansion.

SenseTime's current AIDC operating model already places responsibility on the company for technical architecture, GPU deployment planning, liquid cooling standards and workload enablement.

Candidate Requirements

Must Haves

• 8+ years of infrastructure engineering experience, preferably in HPC, AI data centers, hyperscale cloud or GPU infrastructure.
• Proven experience designing or deploying large scale GPU/HPC environments.
• Strong knowledge of NVIDIA GPU architecture, InfiniBand/RoCE, RDMA and high speed Ethernet.
• Good understanding of data center electrical infrastructure, high density racks, redundancy and PUE.
• Experience with liquid cooling / direct-to-chip cooling for high density GPU environments.
• Knowledge of high performance storage and distributed AI workloads.
• Familiarity with Linux, Kubernetes and/or Slurm.
• Strong ability to coordinate across engineering, facilities, vendors, customers and senior management.
• Fluent English; Cantonese and Mandarin preferred.

Nice-to-Haves

Experience with NVIDIA DGX/HGX, GB200/GB300, Huawei Ascend, heterogeneous GPU clusters, 400G/800G networking, NVMe/NVMe-oF, Lustre/BeeGFS, NCCL optimization, MLPerf/HPL benchmarking, 100kW+ racks, DCIM/BMS/EPMS or >10 MW AIDC deployments.