Quadric has created an innovative general purpose neural processing unit (GPNPU) architecture. Quadric's co-optimized software and hardware is targeted to run neural network (NN) inference workloads in a wide variety of edge and endpoint devices, ranging from battery operated smart-sensor systems to high-performance automotive or autonomous vehicle systems. Unlike other NPUs or neural network accelerators in the industry today that can only accelerate a portion of a machine learning graph, the Quadric GPNPU executes both NN graph code and conventional C++ DSP and control code.

Role
The AI Kernel Engineer in Quadric plays the key role to enable a large number of AI kernels/operators to run efficiently on the Quadric platform. The AI Kernel Engineer at Quadric will [1] develop a highly productive Quadric kernel library for a variety of AI/LLM models; [2] analyze the performance and optimize the kernel for different hardware configurations; This senior technical role demands deep knowledge of hardware architecture, compiler toolchain and optimization techniques.

Responsibilities




Develop AI/LLM kernels/operators on Quadric platform for efficient inference
Optimize the kernel performance for different hardware configurations and workloads
Profile and analyze kernel performance in terms of compute, data and parallelism; identify micro-architecture and software bottlenecks and provide optimization solutions
Optimize kernel C/C++ codes, maximize hardware utilization
Collaborate across related areas of the AI inference stack to support team and business priorities
Make Improvement to Quadric toolchain, compiler and runtime
Provide technical support and documents to customers and developer community

Requirements

Bachelor's or Master's in Computer Science and/or Electric Engineering
5+ years of experience in AI kernel development and optimization
experience with model and kernel inference performance profiling
experience with at least one of the following compute developmen

📌 AI Kernel Engineer (Pune)
🏢 Quadric
📍 Pune