CUDA Engineer

This is a fully remote position, open to applicants in United States.

📋 Description

• Analyze and enhance GPU kernels for optimal performance, efficiency, and hardware usage.

• Utilize profiler metrics like L2 cache hit rate, L2 throughput, occupancy, and other related signals to drive kernel enhancements.

• Examine GPU kernel implementations to pinpoint bottlenecks without needing an extensive background in the underlying algorithms.

• Write, modify, and understand code in C++17, Python, and GPU programming.

• Leverage CUDA, HIP, shader programming, or similar kernel programming knowledge to boost performance results.

• Clearly document optimization choices, including when specific profiler metrics are beneficial or not.

• Submit a resume or relevant technical experience.

• Complete a short technical assessment or provide additional details if required.

⛳️ Requirements

• Available to work a minimum of 20 hours per week.

• Proficient in core C++ features up to C++17.

• Basic knowledge of Python and Git.

• Proficient in at least one GPU programming model, such as CUDA, HIP, Slang, HLSL, GLSL, or similar kernel programming.

• Minimum of 1 year of professional or graduate-level research experience working with GPUs.

• Strong understanding of GPU profiler performance metrics and their application in optimizing kernels.

• Capability to optimize GPU kernels without needing in-depth prior knowledge of every algorithm.

• Experience with CUDA, HIP, CUDA C++ Core Libraries, inline PTX assembly, or tensor core-level optimization is advantageous.

• Experience optimizing kernels for NVIDIA Blackwell hardware is a plus.

• Familiarity with NSight Compute is beneficial.

• Previous experience with GPU hardware companies such as NVIDIA, AMD, or Qualcomm is a plus.

• Contributions to open-source projects related to GPU kernel optimization are a plus.

🏝️ Benefits

• Earn up to $480 for each successful referral.

• Flexible freelance contract opportunity.

• Remote work.