AI/ML Compile Devolpement Engineer

THE ROLE:

We are looking for a dynamic, energetic candidate to join our growing team in AI Group.In this role, the individual will be responsible architecting and defining AI workload models, dataflow, defining block level and system level performance of Neural Processing Unit (NPU), NPU network performance modeling, and performance bottleneck analysis on pre/post silicon platforms. As a member of our dynamic team, you will have the opportunity to shape the future of AI model development

THE PERSON:

• We are looking for a candidate who possesses strong engineering skills to tackle complex challenges on AI model development work.

• You should have experience in optimizing and accelerating CNN/Generative AI models. Person needs excellent cross team collaboration skills to succeed in this role.

• Strong experience in developing ML compiler for efficient network mapping on NPU

• Work with cross-functional teams to optimize various parts of the SW stack - AI Compiler, AI frameworks, device drivers, and firmware.

• Bring up emerging ML models based on CNN, transformers and characterize performance.

• Work on quantization, sparsity, and architecture search methodstooptimizeand enhance the performance, efficiency, and accuracy ofGenerativeAI models.

• Collaborate closely with software engineers, data scientists, and researchers to integrate AI models into software applications and platforms.

KEY RESPONSIBILITIES:

• Research, design, and implement novel methods for efficientCNN, GEN AI models.

• Model optimization method design includingquantization, sparsity, NAS, etc.

• Collaborate with other team members and teams.

• Collaborate with compiler team to develop optimization strategies for the compiler.

PREFERRED EXPERIENCE:

• Experience with deep learning framework, e.g.,Pytorch/ONNX/TensorFlow.

• Experience on model compression, quantization, and end-to-end inference optimization.

• Strong coding skills in C/C++, Python required.

• Experience with any of the following also a plus: LLMs, stable diffusion, NeRF, or text-to-video generation.

• Solid knowledge of AI and ML concepts and techniques. Practical experience applying these concepts to solve real-world problems in the context of research or work experience.

• Understanding the performance implications on AI acceleration of different compute, memory, and communication configurations and hardware and software implementation choices.

• Developing and optimizing code for VLIW processors. Analyzing code for high performance CONV, GEMM and non-linear operators Deep understanding of AI frameworks, preferably ONNX.

ACADEMIC CREDENTIALS:

• Minimum of a BS degree, MS or above preferred.

LOCATION: