Software Development Engineer - C++, Networking

Advanced Micro Devices, Inc•Santa Clara, CA

About The Position

ADVANCE YOUR CAREER. ADVANCE THE WORLD. At AMD, we believe technology has the power to solve the world’s most important challenges. From advancing healthcare and scientific discovery to powering AI and the technologies people rely on every day, innovation at AMD is shaping the future. Whether you’re designing next-gen processors, enabling AI breakthroughs, or bringing leading edge products to market, every role at AMD contributes to something bigger — technology that moves the world forward. Join us and, together, we’ll advance your career. THE ROLE: AMD is looking for a Principal Member of Technical Staff (PMTS) to drive end-to-end delivery of the scale-up networking and GPU software stack for AMD Instinct™ GPU platforms. You will lead architecture, development, bring-up and optimization across scale-up switch networking, GPU firmware, kernel drivers and the ROCm™ software stack, so that rack-scale AI systems deliver top performance, reliability and scalability for large-scale training and inference. You will work closely with silicon, hardware, firmware, switch/networking, system software and validation teams, as well as AMD partners and customers, to bring new Instinct GPU platforms from first silicon to production deployment. This is a senior technical leadership role at the cutting edge of GPU computing and AI infrastructure. THE PERSON: You are a recognized subject matter expert with deep experience across GPU and networking system software, from firmware and drivers up through the user-space stack. You are comfortable working at the hardware/software boundary, can debug complex problems across layers (silicon, firmware, driver, fabric and application), and can set technical direction for multiple teams. You communicate clearly with engineers, architects, executives and customers, and you work well in a highly collaborative, cross-functional environment.

Requirements

Extensive experience in system software development for GPUs, accelerators, NICs or network switches.
Strong background in data center networking: Ethernet (L1–L3), high-speed SerDes, FEC, RDMA/RoCE, congestion control, and switch ASIC programming (e.g., Broadcom SDK, SAI, SONiC).
Familiarity with scale-up interconnects and fabrics such as UALink, Ultra Ethernet, AMD Infinity Fabric™, NVLink or similar.
Hands-on experience with firmware development for SoCs or accelerators, including embedded C/C++, boot flows, power management, security, and BMC/out-of-band management (Redfish, PLDM, MCTP).
Strong Linux kernel and device driver development experience (PCIe, DMA, IOMMU, memory management, interrupt handling).
Knowledge of GPU architecture and GPU software stacks (ROCm/HIP, CUDA, or similar), and of collective communication libraries (RCCL/NCCL, MPI).
Familiarity with deep learning frameworks (e.g., PyTorch, JAX, TensorFlow) and distributed training/inference at scale.
Proven track record of leading new hardware platform bring-up and debugging complex hardware/software interactions using tools such as JTAG, protocol analyzers and logic analyzers.
Proficiency in C/C++ and Python; experience with Go or Rust is a plus.
Experience setting technical direction and influencing across multiple teams and organizations.
Excellent communication, collaboration and leadership skills.
Bachelor’s or Master’s degree in Computer Science, Computer Engineering, Electrical Engineering, or a related discipline; PhD a plus.

Responsibilities

Define and drive the software architecture for scale-up switch fabrics that connect AMD Instinct GPUs within a rack-scale system, including Ethernet-based scale-up fabrics and emerging standards such as UALink and Ultra Ethernet.
Lead switch software integration (e.g., SONiC, SAI, switch SDKs), including link bring-up, SerDes/retimer tuning, FEC/BER characterization, topology discovery, routing, congestion management and telemetry.
Architect and develop GPU firmware components (e.g., system management, power/thermal, security, RAS, and fabric/link management), and drive firmware packaging, update flows (PLDM/Redfish) and BMC integration.
Lead development and optimization of Linux kernel drivers for Instinct GPUs (amdgpu/KFD), including memory management, GPU-to-GPU communication over the scale-up fabric, partitioning, virtualization, error handling and recovery.
Drive enablement and performance of the ROCm stack on scale-up systems, including the runtime, communication libraries (RCCL), and integration with frameworks such as PyTorch, JAX and inference serving engines (e.g., vLLM, SGLang).
Own new-platform bring-up from pre-silicon (emulation/simulation) through first silicon, system integration, and production qualification.
Lead cross-layer debug and root-cause analysis of complex issues spanning silicon, firmware, driver, fabric and application layers.
Define performance, reliability and scalability goals for collective communication and multi-GPU workloads, and drive the optimizations needed to meet them.
Collaborate with silicon architects, hardware, switch, cloud/OEM partners and customers to influence product roadmaps and strategic technical decisions.
Mentor engineers, set technical direction and engineering best practices, and coordinate execution across globally distributed teams.