Principal Software Developer - GPU AI/HPC kernels

Advanced Micro Devices, Inc•Austin, TX

25d•Hybrid

About The Position

At AMD, our mission is to build great products that accelerate next-generation computing experiences—from AI and data centers, to PCs, gaming and embedded systems. Grounded in a culture of innovation and collaboration, we believe real progress comes from bold ideas, human ingenuity and a shared passion to create something extraordinary. When you join AMD, you’ll discover the real differentiator is our culture. We push the limits of innovation to solve the world’s most important challenges—striving for execution excellence, while being direct, humble, collaborative, and inclusive of diverse perspectives. Join us as we shape the future of AI and beyond. Together, we advance your career. THE ROLE: AMD is looking for a principal software developer to join our growing team. As a key contributor you will be part of our ROCm GPU-compute mathematical libraries team working on innovative hardware and software technologies. You will help deliver exceptional performance and feature enhancements via maintainable code development, optimizations/tuning, and collaboration. THE PERSON: The successful candidate would possess excellent software skills in C/C++, Python and has expert understanding of software performance and computer architecture. Must have strong desire and ability to solve problems, and be able to schedule, and work on tasks independently. Should be a great team player with excellent communication skills. THE ROLE: As GPU Software Architect, you will provide technical leadership at the intersection of GPU architecture, multi‑ASIC platform bring‑up, and software enablement for next‑generation GPU products. This is a "Software-First" architecture role: you will reimagine and redefine the end-to-end software libraries lifecycle as it spans across multiple ASICs to create a unified software fabric and process supporting development of software libraries on cutting edge hardware. You will serve as a bridging authority between software architecture and the hardware ecosystem, ensuring that architectural intent translates into working, performant, and scalable solutions for partnerships established with software libraries teams. This role is focused on leading a team focused on new GPUs and new product introductions, with accountability spanning early architecture definition, pre‑silicon modeling, multi‑ASIC bring‑up strategy, and software readiness for emerging platforms. THE PERSON: You are a deeply technical leader who thrives in ambiguous, first‑of‑kind environments. You combine strong architectural intuition with hands‑on experience bringing up complex hardware platforms and enabling software stacks on new silicon. You are comfortable operating across layers — from architecture trade‑offs and interconnect topology down to firmware, drivers, and performance tooling — and you are trusted to make decisions when data is incomplete. You naturally connect and lead teams through technical credibility, clarifying intent, and reducing friction between hardware and software organizations.

Requirements

10+ years professional software development experience
Demonstrated capacity to technically lead and people manage junior to mid-level developers
Proficient in C/C++ & Python programming employing best software design practices
GPU software development or validation involving HIP, CUDA, or OpenCL
Experience with software libraries and API design
Exposure to Matrix/Tensor operations and numerical work
Experience in software performance estimations, optimizations and debugging
Ability to closely interact with technical leads, developers, and test teams to maintain and release production software
Deep experience in GPU, accelerator, or SoC architecture, including memory systems, interconnects, and scalability considerations.
History of technical leadership across distributed, cross‑functional engineering teams.
Strong background in systems software, firmware, drivers, or performance software used to enable new silicon.
Proven experience in hardware/software co‑design, including defining interfaces and debugging cross‑layer issues.
Hands‑on programming experience in C/C++ and Python.
Familiarity with low‑level debugging tools and workflows.
Experience working with performance modeling, simulators, or early validation infrastructure.
Advanced degree in Computer Engineering, Electrical Engineering, Computer Science, or equivalent practical experience.

Nice To Haves

Software emulation to support FP numerical formats is a plus
Applied experience using AI‑assisted coding tools in professional software engineering workflows, including code generation, refactoring, test creation, documentation, and design exploration.
Advanced degrees, such as M.Sc., M.Eng., Ph.D. are preferred

Responsibilities

Develop software in C++, Python, HIP, assembly, and SOTA programming technologies to enable key mathematical operations on GPU
Design GPU computational software libraries for AI, HPC applications
Aid management in planning, and delivering industry-leading software for current and future processors
Supervise small development team
Carry-out performance optimizations and projections for important use-cases to maximize hardware utilization
Support development of programs to sustain seamless performance analysis, and performance/functional test coverage
Identify and help resolve quality issues working closely with libraries development teams and other internal engineering teams
Architecture & Platform Leadership: Provide technical leadership for GPU architecture decisions with direct impact on multi‑ASIC platforms, interconnects, memory systems, and scalability. Translate architectural concepts into concrete platform requirements spanning ASIC, firmware, drivers, and software libraries.
Multi‑ASIC Bring‑Up Strategy: Define and lead bring‑up strategies for new GPU platforms, including strategies spanning multiple ASICs. Partner with silicon, systems, and software teams to identify risks early and drive mitigation plans from pre‑silicon through first silicon.
Hardware–Software Co‑Design: Drive hardware/software interface definition, ensuring architecture choices support and reflect the drive towards performance and quality. Influence firmware, driver, runtime, and performance software design to align with architectural intent.
Early Silicon Enablement: Act as a technical escalation point during early silicon bring‑up, debugging complex cross‑layer issues spanning hardware, firmware, and software. Guide the creation of diagnostics, validation tools, and bring‑up workflows that scale across teams and products.
Cross‑Functional Technical Leadership: Work across architecture, design, verification, drivers, performance libraries, and product teams to ensure alignment. Provide technical mentorship and review, raising the overall effectiveness of teams working on new GPU platforms.
Knowledge Capture & Reuse: Capture lessons learned from new product bring‑up and translate them into reusable architecture patterns, best practices, and documentation.
Leverages AI‑assisted software development tools to accelerate the design, implementation, review, and documentation of complex software libraries. Establishes best practices for responsible use of AI assistance, including validation, review, and traceability of generated code and technical artifacts.