Storia: Base Optimization Stack: From Open Weights to Frontier On-Device Inference Speed — Warptech Lab News