We are an AI-GPU architecture team at Intel, dedicated to leveraging insights from AI workloads to shape future Intel GPU design decisions.
We are looking for passionate individuals to help define and build the next generation of highly efficient GPU architectures for AI. This role offers the opportunity to grow into a performance architect position, contributing to both hardware and software innovations.
Responsibilities:
Conduct workload analysis and performance debugging to identify bottlenecks and drive resolution through hardware and/or software fixes.
Evaluate AI-GPU hardware architecture and influence the product roadmap through a deep understanding of AI algorithms, customer needs, and software frameworks.
Develop and enhance internal performance analysis tools.
Develop highly optimized GPU kernels and software stacks, collaborating with partner teams to deliver high-performance solutions.
Provide a comprehensive view of solutions and support both pre- and post-silicon activities.
Collaborate with experts to analyse next-generation requirements and guide research and academic partnerships.
Minimum Qualifications:
MS in Computer Engineering, Computer Science, Electrical Engineering, or Mathematics.
2-3 years of experience in GPU/CPU architecture for AI workloads.
Proficient in Python, C/C++.
Good data analysis and presentation skills.
Preferred Qualifications:
Knowledge of AI and deep learning, including Large Language Models (LLM) and Stable Diffusion.
Experience in performance analysis/performance debugging.
Experience in building analytical and/or simulation-based performance models.
Knowledge/experience in CPU, GPU, or memory design/architecture, and/or microarchitecture/RTL/design/process technologies.
An aptitude to learn new things quickly in this fast-changing domain and flexibility to work on any mix of the above roles as needed.
This role is ideal for candidates who are eager to learn about AI hardware, possess strong collaboration skills, and can thrive in a fast-paced hardware/software development environment. Requirements listed would be obtained through a combination of industry-relevant job experience, internship experiences, and/or schoolwork/classes/research.
Work Model for this Role
This role will be eligible for our hybrid work model which allows employees to split their time between working on-site at their assigned Intel site and off-site. * Job posting details (such as work model, location or time type) are subject to change.