Embedded Files

VAERAX FORGE LLC (Daytona Beach, FL) — Service-Disabled Veteran-Owned Small Business (SDVOSB) developing a native C++20/CUDA 13.2 model compression compiler, zero-Python inference SDK, and 3-slot Cognitive OS runtime for constrained edge GPUs (NVIDIA Jetson Orin Nano 8GB on JetPack 7.2.1 / L4T R39.2.1 and discrete x86_64 GPUs). Our 15-mode Rate-Distortion Optimization compiler compresses US-provenanced 3.2B to 14.7B language and reasoning models (Meta Llama-3.2, NVIDIA Llama-3.1-Nemotron-8B, Microsoft Phi-4) by 4.65x to 6.94x—enabling a 14.7B reasoner (4.22 GB at 4.39 tok/s) or a 3-model co-resident swarm (3.2B Router + 8.0B Specialist + 3.8B Reasoner = 15.0B combined parameters in 5.80 GB) to execute 100% offline inside a single 8GB Jetson Orin Nano.

Page updated
Report abuse