We are building a next-generation storage platform for AI infrastructure that combines high-performance flash, accelerator technologies, and advanced storage software, with the goal of delivering a breakthrough step-function improvement in cost, power efficiency, density, and scalability for AI-era data-center storage. We are seeking a R&D Lead & Inference Storage System Architect to define and drive the end-to-end architecture of this platform, from data-center deployment models down to node-level HW/SW partitioning. This role will also contribute to strengthening Sandisk’s broader data-center infrastructure architecture knowledge across the Architecture organization, working collaboratively with existing domain experts and helping advance our system-level deployment thinking.
Responsibilities:
- Drive the architecture of a groundbreaking data-center storage system targeting step-change improvements in cost, power, density, and scalability
- Define the overall architecture of the storage platform and its deployment model within large-scale AI infrastructure environments
- Analyze data-center deployment models, including hyperscaler environments, AI training and inference clusters, and disaggregated storage approaches
- Define system data flows, metadata flows, and identify performance choke points
- Architect the storage rack and tray topology
- Define the storage node architecture, including SoC/DPU selection, PCIe/NVMe topology, DRAM architecture, and NIC integration
- Drive system partitioning decisions across hardware, firmware, and software components
- Define placement of RAID, compression, and data services
- Lead bottleneck analysis and scalability modeling at node and rack levels, including multi-node and large-scale system behavior
- Own system-level power and performance modeling
- Engage directly with customers and ecosystem partners to align architecture with real deployment needs
- Guide cross-company technical collaboration and ensure architectural alignment across internal and external contributors