Low-Level AI Engineer · Spain
Running AI where it belongs — on the edge.
C/C++ CUDA ROCm/HIP Vulkan WebGPU/WGSL llama.cpp
QNN NNAPI/ANE Go Dart/Flutter TypeScript
Specialized in local AI inference at the hardware level — NPU optimization, GPU compute (CUDA/ROCm/Vulkan/WebGPU), and low-level inference engines.
Building NPU Universe — testing real NPU performance with LLMs across devices, from phones to edge hardware.
| Project | Description | Stack |
|---|---|---|
| NPU Universe | NPU performance testing platform with real LLMs and offline inference | Go · Dart · C++ · TypeScript |
| TheRock | The HIP Environment and ROCm Kit — lightweight open source build system for HIP and ROCm | Python · Shell · C++ · CMake |
| npu-optimize | Hardware-aware CLI that detects NPUs/GPUs, finds compatible GGUF models, and recommends optimal llama.cpp inference configs | Go |
| llama.cpp | Custom fork with Windows build fixes and NPU optimizations | C++ · CUDA · Metal |
Flutter · Android NDK · React/TypeScript · Node.js · Electron · Tauri