Software Engineer - Gpu Performance Job in Cynlr - Cybernetics H.i.v.e

Job Title: Software Engineer GPU Performance

Location: Bengaluru

Overview:

We are looking for a highly skilled Software Engineer GPU Performance with a deep understanding of CUDA, GPU hardware architecture, and low-level performance optimization. The ideal candidate will have hands-on experience building high-performance GPU-based pipelines, optimizing time-continuous kernels, and dynamically managing processing loads between the CPU and GPU.

Key Responsibilities:

• Utilize low-level CUDA APIs to implement and optimize GPU kernels and memory management strategies.
• Design and optimize pipelined image processing frameworks, ensuring seamless multi-block function execution and inter-block communication.
• Conduct low-level GPU performance analysis and optimizations using tools like:
  • NVIDIA Nsight Compute
  • NVIDIA Visual Profiler
  • NVIDIA Graphics Developer Tools
• Optimize CUDA cores and kernels for maximum throughput, particularly in time-continuous processing scenarios.
• Implement dynamic load balancing between GPU kernels and processing functions.
• Design interleaved execution strategies between CPU and GPU, including real-time GPU control flow modifications from the CPU.
• Use NVIDIA Direct technologies for direct memory access from PCIe, USB, and display hardware, bypassing CPU intervention.
• Build systems to visualize GPU memory for debugging without requiring CPU transfers.
• Contribute to the design and optimization of foundational neural networks, including mathematical modeling of time-weighted kernels.
• Stay up to date with emerging GPU tools and platforms; exposure to NVIDIA Omniverse is a plus.

Required Skills & Qualifications:

• Strong proficiency in C/C++.
• In-depth experience with low-level CUDA programming.
• Proficiency with Visual Studio toolchain and related debugging tools.
• Solid understanding of GPU hardware architecture and system-level performance tuning.
• Hands-on experience with GPU memory management, kernel interleaving, and CPU-GPU orchestration.
• Strong problem-solving skills and the ability to write clean, efficient, and maintainable code.
• Experience in neural network architecture design and low-level performance optimization is highly desirable.
• Exposure to Omniverse, real-time rendering, or simulation platforms is a bonus.