Key responsibilities :-
Client requirement discovery: lead workshops with client CIO/CTO and infrastructure teams to capture workloads, density profiles, power and cooling needs, network, interconnection and sovereignty requirements, and growth phasing
HPC/AI solution design: architect end-to-end solutions for AI training and inference environments
GPU compute cluster sizing and topology, storage throughput, and high-performance network fabrics mapped onto colocation space, power and cooling.
Colocation engineering design: develop space and rack layouts, power topologies (N / N+1 / 2N), and cooling strategies including rear-door heat exchangers and direct-to-chip liquid cooling for 30–130 kW racks.– GPU & InfiniBand architecture: define rack-scale and pod-scale designs for NVIDIA HGX/DGX-class systems, including InfiniBand and RoCE fabric topologies (rail-optimised, fat-tree), cable plant planning and density zoning.
Proposals & RFPs: author solution documents, BOMs, technical proposals and RFP/RFI responses; present and defend designs to client executive and engineering audiences.
Techno-commercial support: partner with the commercial team on capacity pricing, fit-out scope splits (shell / powered shell / fully fitted), phasing and SLAs.
Build alignment: work with the DC Infrastructure arm to ensure designs are buildable, costed and aligned to campus master plans.
Strategic support: lead technical due diligence for anchor tenants and JV partners; mentor junior solution architects and build the solution engineering playbook.
Essential qualifications & experience (must-have)
10–15 years in data centre solution architecture, pre-sales engineering or design consulting, with a minimum of 3 years dedicated to HPC/AI solution design – designing, dimensioning and deploying GPU compute environments for AI training and inference workloads.
Hands-on GPU and InfiniBand experience: NVIDIA HGX/DGX-class cluster architecture, InfiniBand (NDR/HDR) and RoCE fabric design, NVLink/N