Rack Scale Serviceability & Telemetry Architect

Advanced Micro Devices, Inc•Austin, TX

21d•Hybrid

About The Position

AMD's Data Center GPU Systems Architecture team defines next-generation AMD Instinct platforms and complete rack-scale solutions for hyperscale AI and HPC deployments. We work across silicon, GPU system firmware, server and board architecture, BMC/platform firmware, management software, security, validation, manufacturing, and ecosystem partners to turn product strategy into deployable, serviceable, production-ready platforms. AMD is seeking a Principal Member of Technical Staff (PMTS) to own the architecture for rack-scale serviceability and telemetry across AMD Instinct product lines and complete rack-scale solutions. This is a highly visible technical leadership role responsible for defining the end-to-end manageability, observability, and serviceability architecture spanning node, chassis/tray, rack, and fleet domains. You will drive the strategy, architecture, execution, and delivery of standards-based solutions for inventory, discovery, health monitoring, telemetry, eventing, diagnostics, firmware lifecycle management, and field service workflows across the full AMD rack-scale stack. In this role, you will independently own a critical cross-product architecture area and drive alignment across GPU/SoC architecture, server/platform architecture, BIOS/UEFI, BMC and embedded software, security, RAS, validation, ODM/OEM partners, and customer-facing teams. The role spans early concept definition through bring-up, validation, deployment, and post-launch improvement.

Requirements

Expert level experiences in platform architecture, system management, BMC/embedded firmware, server manageability, or adjacent domains, including significant time in architect or technical leadership roles.
Proven experience defining serviceability/manageability architecture for servers, accelerators, storage, networking, or rack-scale infrastructure in datacenter, cloud, AI, or HPC environments.
Deep knowledge of DMTF Redfish, including schema design, OEM extension strategy, eventing, update service, and telemetry concepts such as MetricReportDefinition/Metric Reports; strong understanding of PLDM/MCTP for platform inventory, monitoring, control, and update workflows.
Strong hands-on experience with OpenBMC, including Yocto/OpenEmbedded, D-Bus, systemd, bmcweb/Redfish, phosphor services, firmware update flows, sensor frameworks, and log/event handling.
Experience with embedded Linux, ARM-based BMC SoCs, U-Boot, Linux kernel/device driver concepts, device tree, and low-level interfaces such as I2C/I3C, SPI, UART, GPIO, SMBus/PMBus, and related platform-management buses.
Strong understanding of server/platform RAS and serviceability features such as health monitoring, error logging, crashdump, diagnostics, inventory/FRU management, and remote recovery.
Experience with secure manageability architectures, including secure boot, root of trust, attestation, firmware signing, SPDM, and protection of out-of-band management paths.
Experience creating architecture specifications, product requirements, conformance plans, validation strategies, and design reviews that drive execution across multiple internal teams and external partners.
Strong programming and scripting background in C/C++, Python, and shell, with the ability to debug across firmware, hardware, and system software boundaries.
Strong written and verbal communication skills with proven ability to influence senior engineering leadership, customers, and strategic partners.
Bachelor’s or Master’s degree in Computer Science, Computer Engineering, Electrical Engineering, or a related technical field. Advanced degree preferred.

Nice To Haves

Experience with large-scale telemetry or observability pipelines, metrics consumers, or fleet operations tooling is strongly preferred.
Experience with AMD server or GPU platforms, AI/HPC system design, liquid cooling/power/thermal infrastructure, or OCP-aligned rack architectures is a plus.

Responsibilities

Define and own the end-to-end rack-scale serviceability and telemetry architecture for AMD Instinct-based solutions, spanning node BMC, chassis/rack management, service processors/controllers, management network, and fleet-level observability integration.
Define the standards strategy and interface architecture using DMTF Redfish, PLDM, MCTP, and related specifications, maximizing standards compliance while establishing AMD/OEM extensions only where required.
Drive OpenBMC-based architecture and implementation direction for BMC and rack management controllers, including D-Bus object models, bmcweb/Redfish requirements, sensor and FRU inventory models, logging, eventing, firmware update, and debug workflows.
Architect telemetry frameworks for health, power, thermal, inventory, error, utilization, and service data. Define schemas, metric taxonomies, triggers, event models, aggregation, retention, and reporting strategies required for at-scale observability and automated service operations.
Define platform serviceability flows covering discovery, inventory correlation, fault isolation, diagnostics, crashdump and error capture, remote recovery, FRU replacement, firmware/driver update orchestration, and return-to-service procedures.
Partner with GPU/SoC architects, board and system architects, firmware and software teams, security/RAS, validation, manufacturing, and customer engineering to translate requirements into production-ready architecture and deliverables.
Work closely with ODM/OEMs and ecosystem partners to review designs, close gaps, guide implementation trade-offs, and deliver robust reference solutions and customer platforms on schedule.
Drive validation and conformance strategy for manageability and telemetry, including interoperability, Redfish/PLDM compliance, fault injection, service workflow validation, scale testing, and field debug methodology.
Influence future AMD Instinct platform roadmaps using insights from bring-up, partner integrations, deployment learnings, and telemetry-driven data.
Represent AMD in relevant standards and open-source communities, including DMTF and OpenBMC forums, and guide upstream/downstream strategy where appropriate.
Mentor engineers and architects across the organization and serve as the senior technical point of contact for rack-scale serviceability and telemetry.