· Performance Optimization: Profile and optimize existing C/C and CUDA code to achieve real-time performance for a comprehensive software ISP pipeline.
· CUDA Implementation: Translate complex image processing algorithms into highly parallelized CUDA kernels, optimizing memory access and execution configurations.
· Rapid Prototyping: Utilize Python scripting to develop and validate Proof of Concepts (PoCs) for quick algorithm evaluation.
· Code Quality: Deliver clean, maintainable, and well-documented C/C and CUDA code.
Required Qualifications
· Experience: 5 years of hands-on experience in CUDA optimization targeting NVIDIA server-range or desktop-range GPUs.
· Programming: 5 years of robust coding experience in C/C , plus proficiency in Python scripting (and libraries like scipy, numpy, matplotlib) for PoCs.
· GPU Architecture: Deep understanding of NVIDIA GPU architectures, execution models, and profiling tools (Nsight Systems/Compute, nvprof).
· Libraries: Strong knowledge of OpenCV, specifically working with and integrating its CUDA-optimized modules.