Description

Port and validate PyTorch-based Large Language Models on custom AI processors using CUDA SDK APIs.

You will own the integration between CUDA code and PyTorch models, ensuring optimal performance on specific hardware.

Responsibilities

• Port and validate PyTorch-based LLMs on company-specific AI processors.
• Extend, modify, and replace CUDA code to optimize performance.
• Debug and troubleshoot complex integration issues between CUDA and PyTorch.
• Develop validation frameworks for LLM performance analysis.
• Collaborate with teams to ensure seamless hardware-software integration.

Required Skills

• 5+ years of experience in machine learning engineering or related software roles.
• Proficiency in C++ and Python programming.
• Hands-on experience with CUDA programming and CUDA SDK APIs.
• Strong experience with the PyTorch framework.
• In-depth knowledge of Large Language Models (LLMs) and deep learning architectures.
• Proven ability to debug and troubleshoot complex software and kernel-level issues.
• Bachelor's or Master's degree in Computer Science, Electrical Engineering, or a related field.

Preferred Skills

• Experience with AI processor architecture and design.
• Knowledge of TensorFlow or other deep learning frameworks.
• Experience profiling and tuning GPU kernels for speed, memory usage, and scalability.

Key Skills

Education

Bachelor's or Master's degrees