Sipp: a local-first runtime for Hybrid AI Applications

Over the past few months, I had the opportunity to contribute to llama.cpp’s WebGPU backend, helping push it from isolated operator support toward a more complete and reliable path for browser-based a...