We are seeking a skilled and innovative Senior Kernel Engineer to help build and optimize our clients cutting-edge Software Development Kit (SDK). Our clients SDK is designed to deliver industry-leading performance and power efficiency for our clients unique at-memory AI inference chips. By enabling customers to compile AI models directly for their architecture, and providing advanced tools for performance analysis and optimization, were setting a new standard in AI inference technology.

At the core of our SDK is the kernel library, which maximizes hardware efficiency through fast computations, flexible parallel workloads, and highly configurable data-flow options. This role requires a deep understanding of our architectures strengths and constraints, and how the kernel libraries features enable seamless and performant compilations.

Key Responsibilities:

• Develop Neural Network Compute Kernels: Design and implement efficient and flexible kernels tailored for our chip families.
• Enhance Kernel Abstractions: Define and improve abstractions in our kernel library to streamline development and accelerate kernel performance.
• Optimize Performance: Conduct in-depth analysis and optimization of individual kernel and full-network implementations.
• Collaborate Cross-Functionally: Work closely with the compiler and physical allocation teams to deliver efficient, push-button compilation workflows for AI models.

Required Qualifications:

• Bachelors or advanced degree in Computer Science, Engineering, Mathematics, Physics, or a related field.
• Proficiency in Python, C/C++, and software design principles.
• Proven ability to work independently on complex, tightly constrained technical challenges.
• Demonstrated experience leading technical initiatives and collaborating with cross-functional teams.
• Strong interest and ability to balance high-level architecture considerations with detailed, low-level technical work.
• Expertise in low-level and parallel optimization techniques, including assembly language, GPU shaders, SIMD, CUDA, or AI inference accelerator kernels.

Preferred Qualifications:

• Familiarity with spatial architectures and at-memory compute technologies.
• Knowledge of AI algorithms and neural network principles.
• Advanced mathematical skills and a passion for solving complex problems.
• Enjoy tackling challenging intellectual problems, such as tricky math puzzles or logic-based tests.

If you're excited about pushing the boundaries of performance and innovation in AI inference, we'd love to hear from you!