Hardware System Architect Lead, GPU Platform

About Think

Think is revolutionizing compute efficiency globally by integrating hardware and software development under one roof since its foundation in Riyadh in 2025. The company addresses the underutilization of silicon caused by software limitations, aiming to deliver the most efficient AI infrastructure, enabling accessible, scalable, and uncompromised frontier AI operations.

Think’s core principles emphasize ownership, agility, impact, and mastery, expecting team members to own outcomes, move swiftly, focus on impactful results, and pursue excellence in solving complex challenges.

Role Overview

The Hardware System Architect Lead will be responsible for the complete architecture of Think AI's multi-accelerator server platforms, which include high-density GPU systems supporting NVIDIA, AMD, and Intel accelerators with one-to-one GPU-to-NIC affinity. This role spans from initial design through manufacturing release at the Riyadh facility and ongoing platform evolution. The incumbent will transition platform design from partners to in-house teams, overseeing quality assurance at every level.

Key Responsibilities

• Lead end-to-end system architecture, including CPU platform choice, PCIe Gen5/Gen6 topologies, lane budgeting, switch selection, and maintaining GPU-to-NIC affinity.
• Engineer support for GPUDirect RDMA and peer-to-peer communication across diverse accelerators, ensuring compliance with ACS, ATS, IOMMU, and P2P standards per silicon vendor.
• Develop and execute signal integrity strategies for complex board components in collaboration with layout and SI engineers.
• Direct power and thermal solutions for dense, high-TDP accelerator arrays encompassing CRPS power delivery, airflow guidance, and liquid-cooling readiness with mechanical/thermal experts.
• Define BIOS/UEFI and BMC requirements and coordinate with software teams on hardware-orchestration integration.
• Create platform specifications and oversee partner design execution; conduct thorough schematic, layout, and BOM reviews.
• Manage ODM and JDM partner relationships, guiding platforms through EVT, DVT, and PVT phases toward production in Riyadh.
• Steer the shift from outsourcing to internal board design with progressive ownership over schematics, layout, and signal/power integrity.
• Design and supervise platform validation and quality assurance strategies, supporting comprehensive testing and qualification.
• Implement Design for Test and Manufacturing strategies ensuring thorough test coverage and production yields at the manufacturing site.
• Lead failure analysis from initial power-up through volume production, managing corrective actions and collaboration between manufacturing and customers.
• Develop automation tools in Python and Bash integrated with telemetry systems for scalable testing and validation.
• Build and mentor the hardware platform team, covering board design, signal integrity, firmware, and validation disciplines.

Candidate Profile

• Minimum 12 years experience in hardware or system architecture with a focus on GPU-accelerated or dense accelerator server platforms within vendors, OEMs, ODMs, or hyperscalers.
• Proven record of delivering at least one high-density multi-GPU server platform from concept through production.
• In-depth expertise in PCIe technology including switch topologies, Gen5/Gen6 signal integrity, retimers, resource allocation, and enumeration.
• Comprehensive knowledge of RDMA NICs, GPUDirect, NUMA, IOMMU, power and thermal challenges in dense server designs.
• Proficient in schematic reading, layout review leadership, and laboratory debugging using scopes and PCIe analyzers.
• Hands-on experience in platform bring-up, validation, failure analysis, factory debugging and production ramp-up.
• Experience drafting validation plans, qualification matrices, and leading root cause investigations supported by Python and Bash automation.
• Strong command of English.

Desirable Qualifications

• Practical experience in schematic design for server motherboards, risers, or PCIe switch-related boards.
• Experience integrating multiple accelerator vendors on a single platform.
• Familiarity with open interconnect standards such as UALink, Ultra Ethernet, NVLink, and CXL.
• Knowledge of advanced baseboards (HGX, OAM, UBB), rack integration (OCP, ORv3), and liquid cooling systems.
• Expertise in DDR5 memory qualification, margin analysis across DIMM vendors, and advanced reliability testing methods (HALT, HASS).
• Experience with rack-level integration and testing.
• Background working in an early-to-growth stage deep tech startup environment.

What Think Offers

• Opportunity to join a pioneering deep tech company in the region with in-house foundational technology development.
• Significant ownership and real impact at an early stage of company growth.
• Competitive compensation for early-stage involvement.
• Close collaboration within a small, senior team.
• Challenging problems involving hardware, systems, and AI at scale.