Remote CUDA Kernel & GPU Performance Expert

24-Mag Llc; Northern, New York (KY, NY); Hybrid; USD 83,000 - 138,000; Full time

Job summary

24-MAG LLC invites experienced CUDA engineers for a fully remote contracting engagement focused on CUDA, C++, GLSL, and WebGPU. You will analyse GPU kernels, profile workloads, and push performance across modern GPUs.

Ideal candidates have a proven track record in HPC, memory and compute analysis, and strong communication across remote teams; no AI-training experience required. Compensation is hourly, with rigorous benchmarking and documentation of measurable improvements for stakeholders across

Qualifications

• Demonstrated CUDA programming expertise.
• Strong GPU kernel performance optimization track record.
• Advanced C++ development experience.
• Experience in high-performance computing environments.
• Hands-on GLSL and WebGPU experience.
• Proficient with GPU profiling tools (Nsight, Visual Profiler).
• Ability to analyze memory behavior and hardware utilization.
• Excellent written and verbal technical communication.
• Comfortable collaborating in remote, cross-disciplinary teams.
• No prior AI-training experience required.

Responsibilities

• Analyse and optimise GPU kernels using CUDA.
• Identify performance bottlenecks affecting throughput.
• Improve kernel efficiency across modern GPU hardware.
• Apply profiling-based optimisation techniques.
• Validate performance gains with benchmarks.
• Profile workloads with Nsight or Visual Profiler.
• Diagnose memory, compute, occupancy, and bottlenecks.
• Evaluate performance across hardware generations.
• Refactor CUDA and C++ code for maintainability.
• Develop GLSL shader logic and WebGPU workflows.
• Document optimisation strategies and results.
• Communicate GPU behavior to stakeholders.

Skills

CUDA programming

GPU kernel optimization

Advanced C++

HPC environments

GLSL

WebGPU

GPU profiling

Memory analysis

Tools

Nsight

Visual Profiler

Job description

24-MAG LLC invites experienced CUDA engineers for a fully remote contracting engagement focused on CUDA, C++, GLSL, and WebGPU. You will analyse GPU kernels, profile workloads, and push performance across modern GPUs.

Ideal candidates have a proven track record in HPC, memory and compute analysis, and strong communication across remote teams; no AI-training experience required. Compensation is hourly, with rigorous benchmarking and documentation of measurable improvements for stakeholders across