Head of Platform Software
D-MatrixSanta ClaraRemote
Needs a free Superjobs account. You'll finish on D-Matrix's Ashby page.
Skill score
Summary
The Head of Platform Software owns the physical platform layer for rack-scale AI accelerator systems, covering device and link discovery, health, fault detection, readiness gating, firmware and BMC integration, and bring-up tooling. This leader also builds and manages the platform software team and publishes a versioned interface to the fabric manager, runtime, and schedulers, suiting a senior systems leader with deep infrastructure and team-building experience.
- Function
- Development
- Estimated salary
- Est. $263k–395k/yr
- Estimated for a director-level role, ~12 yrs experience, United States. Not published by the employer.
- Work mode and location
- Remote, Santa Clara
- Experience
- 12+ yrs
- Employment type
- Full-time
Job description
Overview
The platform software layer sits between hardware bring-up and the workload stack. It determines whether a rack-scale AI accelerator system powers on, joins its fabric, passes validation, and stays healthy in the field. As the senior leader of platform software, you will own the physical platform layer: device and link discovery, health, fault detection, readiness gating, firmware and BMC integration, and bring-up and fleet tooling. You will build the team that delivers it and publish a clean, versioned interface to the fabric manager, runtime, and schedulers above it. You will partner closely with hardware, runtime, diagnostics, validation, deployment, operations, and product teams.
Roles and responsibilities
- Own node, rack, and cluster bring-up software, including discovery, provisioning, configuration, and readiness gating before any workload traffic, in partnership with the fabric software team.
- Define and maintain a versioned interface between the platform layer and the workload stack (fabric manager, runtime, and schedulers), with explicit state transitions and events pushed to consumers rather than inferred.
- Own device and link discovery and health, and publish normalized state and events for devices and links across their lifecycle.
- Drive fault detection and classification, including severity-based handling, escalation paths, and hand-offs to recovery owners and data center operations.
- Own firmware, BMC, and low-level platform software integration, packaging, and update flows.
- Deliver health telemetry, diagnostics integration, and fleet monitoring for deployed clusters, and integrate with orchestration so the platform layer owns device and fabric readiness.
- Build tooling and recipes to turn up, stabilize, and hand off initial clusters to production operations at a defined readiness threshold, with feedback loops to hardware, diagnostics, manufacturing, and field teams.
- Hire and lead the platform software team, owning technical direction, milestones, quality, and on-call, and turn architecture into owned, dated deliverables.
Skills required
Top skills
- Systems software
- Rack-scale infrastructure
- Hardware bring-up
- Firmware update frameworks
All skills (12)
- BMC and out-of-band management
- Redfish
- OpenBMC
- PLDM
- Fault handling
- Fleet management
- Engineering team leadership
- Interface and contract design
- Kubernetes
- Device plugins
- High-speed interconnect
- Accelerator or GPU infrastructure
Qualifications
- Bachelor’s degree in Computer Science, Electrical or Computer Engineering, or a related field, or equivalent practical experience.
- 12+ years in systems software, including substantial experience with rack-scale, cluster-scale, or large server infrastructure.
- Track record of taking a platform or infrastructure program from design to shipped software with milestones.
- Depth in several of the following: hardware bring-up, device and firmware interfaces, BMC and out-of-band management, fleet management, and fault handling.
- Hands-on experience with BMC and out-of-band management standards (Redfish, OpenBMC, PLDM) and firmware update frameworks.
- Experience building and leading engineering teams, including hiring, growing leaders, and managing cross-functional dependencies.
- Strong judgment in interface and contract design between systems layers.
Nice to have
- Master’s degree in a related field.
- Experience with accelerator or GPU infrastructure, including taking a new hardware platform from early silicon through first customer deployments.
- Experience with high-speed interconnect and fabric software such as PCIe, NVLink-class, CXL, RDMA, InfiniBand, or Ethernet-based fabrics
- Experience with Kubernetes, device plugins, and cluster lifecycle tooling.
Company info
About D-Matrix
d-Matrix makes memory-centric compute hardware for generative AI inference, aiming for ultra-low latency and greater energy efficiency in scalable AI infrastructure. The sources do not state specifically who its customers are beyond general AI infrastructure use.