Machine Learning Kernels Engineer

We are seeking a

Machine Learning Kernels

Engineer to join a well-established and funded startup who are developing a ground-breaking at-memory AI chip architecture. This approach addresses the accuracy and efficiency challenges in running AI inference workloads.

You'll join a team who's mission is to build implementations of real-world neural network kernels specialized for the company's unique hardware architecture - maximizing compute and communication throughput.

Responsibilities: Design and implement low-level kernels for various neural network operations. Work with Compiler team to design configuration APIs for these kernels. Collaborate with architects and compiler engineers to create performance optimizations for future generations. Design overall computation strategies for multi-kernel, multichip neural network implementations.

Requirements: MS or PhD in CS, Math, Physics or related field. Expertise in modern C++ and experience utilizing it in code generation and low-level compute optimizations. Strong knowledge of basic NN operator algorithms (convolutions, transformers, RNNs, etc.) Python experience and strong mathematical skills. Experience working on other AI Accelerators is a huge plus!