During June and July 2025, Etasnadi developed and integrated Vulkan-backed convolution operations for both 1D and 2D workloads in the ggml-org/llama.cpp and Mintplex-Labs/whisper.cpp repositories. Leveraging C++, GLSL, and Vulkan, Etasnadi designed new shader programs and pipeline configurations to accelerate deep learning inference on GPUs, focusing on both transpose and direct GEMM convolution paths. The work included comprehensive testing, performance validation, and cross-platform optimizations, addressing compilation issues and ensuring hardware compatibility. These contributions enabled faster, scalable model deployment and improved inference efficiency, demonstrating depth in GPU programming, shader development, and performance optimization within complex ML frameworks.