A high-precision, hardware-aware router built on Phi-4 Mini (3.8B). It acts as a dispatcher for local AI setups, automatically deciding whether a prompt should run on edge hardware (like a Jetson Nano), a local GPU, or the cloud based on task complexity.