Senior Product Manager – ROCm & AI​/ML Inference Software

What You Do At AMD Changes Everything

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.

AMD is seeking a Senior Product Manager to drive strategy and execution for ROCm, AMD's open-source GPU software stack, with a specific focus on large-scale model inference on AMD Instinct™ and Radeon™ hardware. This is a key, central role sitting at the intersection of the open-source community, high-performance computing, and production AI deployment.

You will join a small but focused team of product managers whose shared remit is inference at scale and the frameworks that enable it. You will influence engineering roadmaps, represent AMD in open-source communities and at conferences, and connect the needs of AMD's most strategic customers with the direction of the open-source ecosystem. AMD's customers include some of the organizations deploying and serving the largest language models in the world.

You are a technically fluent product leader with deep roots in GPU computing, open-source software, and AI inference infrastructure. You thrive on moving fast in a community-driven environment while also navigating the nuanced requirements of enterprise-scale customers. You are as comfortable reading a Git Hub issue thread as you are presenting a roadmap to a VP of Engineering at a hyperscaler.

Product Strategy & Roadmap

• Own the product roadmap for ROCm's inference software capabilities, including integrations with key frameworks like PyTorch and JAX, serving runtimes like vLLM and SGLang, and the libraries and profiling tools that make inference workloads perform well on AMD hardware.
• Define and communicate a coherent strategy for how AMD software enables production inference workloads, covering the full range from single-GPU to rack-scale deployments, with a focus on developer experience, performance portability, and competitive standing.
• Identify when emerging OSS model architectures, serving runtimes, or inference techniques require an AMD software response and drive those requirements into the roadmap.
• Bring a strategic lens to inference by continuously evaluating the latest research, technology trends, and evolving deployment needs, ensuring the organization stays ahead of future inference requirements and translates market signals into actionable product strategy.

Open-Source Community Engagement

• Serve as AMD's active presence in the open-source AI/ML community: monitor Git Hub repositories, Discord servers, developer blogs, and academic papers to track emerging trends, pain points, and opportunities.
• Own and communicate AMD's open-source software roadmap, publishing updates across key community projects to build awareness, solicit feedback from model developers and researchers, and signal AMD's long-term direction to the open-source ecosystem.
• Build and maintain relationships with key OSS maintainers, foundation working groups, and community contributors whose work shapes how AMD hardware is perceived and adopted.
• Represent AMD at major conferences such as SC, PyTorch Conference, and MLSys; write and review technical blog posts and community announcements on the AMD ROCm blog.

Engineering Partnership & Execution

• Partner with engineering leads across inference libraries, kernel, runtime, and serving integration teams to translate developer and customer needs into prioritized engineering work.
• Drive release planning across rapid release cadences with a CI/CD-first mindset and strong attention to compatibility, regression, and ecosystem…