Summary
Bernard Gingold is an NPU compiler engineer with 11 years’ experience specializing in low-level C/C++ and Fortran performance optimization for AI accelerators, DSPs and multimedia engines. He has a proven track record porting and adapting compilers and writing code generators for proprietary NPUs, and has applied deep profiling expertise (Intel VTune, Linux perf, PMCs) to tune AVX/AVX512 and RISC-V vectorized kernels. Bernard’s work spans product-grade projects at GlobalLogic, Samsung and Nokia, where he combined microarchitecture collaboration with hands-on assembly- and intrinsic-level optimization. He also brings experience converting legacy Fortran to optimized C with OpenMP and intrinsics, and has proposed novel image-analysis approaches using temporal/spatial RGB integration for plant-disease detection. Based in Poland, he pairs systems-level rigor with practical tooling knowledge across compilers, profilers and debuggers, often finding gains that aren’t visible at the source level.
11 years of coding experience
7 years of employment as a software developer
Polish, English, Russian