TheOrc.devOregon Radio Company LLC

NVIDIA technology / Oregon Radio Company

Local AI.
Practical tools.
Built in Oregon.

We use NVIDIA CUDA to run local AI in TheOrc. We are also exploring how GPU-assisted analysis could support our radio and facilities software.

Oregon Radio Company LLC is a member of the NVIDIA Inception program.

using today
CUDA · TheOrc
exploring
NVIDIA JetPack SDK, NVIDIA TensorRT, NVIDIA NIM, and NVIDIA NeMo Agent Toolkit
our focus
Local AI, embedded radio, and facilities operations

Current technology use and future work are described separately below. Updated November 8, 2026.

NVIDIA Inception Program member badge

Explore: TheOrc · OrcSDR · OrcNode · FORC Inspection

TheOrc

CUDA in use

AI that runs on your own computer.

TheOrc is our open-source, local-first AI workbench for developers and researchers. It brings together local models, specialist agents, distributed workers, and local LoRA adapter training.

Its native inference path uses LLamaSharp and llama.cpp to run GGUF models with NVIDIA CUDA acceleration. GPU-memory admission controls help decide which work can fit on a GPU. HIVE MIND workers distribute jobs across machines.

Context Fabric adds persistent document memory with source citations. ORC ACADEMY supports local adapter training. The aim is to keep more of the models, documents, and computing under the user's control.

Explore TheOrc on GitHub →

OrcSDR

Companion analysis under consideration

A standalone radio first.

OrcSDR turns ESP32-P4 hardware and an RTL-SDR receiver into a touchscreen radio with live spectrum displays and protocol decoding. Public firmware is available; the integrated appliance remains in development.

The current receiver runs on the microcontroller, without NVIDIA runtime acceleration. We are considering JetPack SDK and TensorRT for a future companion that could assist with RF analysis and signal identification.

See OrcSDR →

OrcNode

In development

More receivers. A clearer view of RF activity.

OrcNode is our developing platform for distributed RF monitoring. Its ESP32-P4 prototypes combine multiple RTL-SDR receivers, spectrum scanning, local change detection, evidence capture, and network reporting.

We are exploring a JetPack SDK and TensorRT companion for GPU-assisted analysis. Signal-family recognition and uncertainty-aware alerts are research directions; an integrated live RF classifier and representative field performance still need to be demonstrated.

Explore the embedded hardware →

FORC Inspection

AI features planned

Less setup work for facilities teams.

FORC Inspection, also featured here as FORCPM, is our facilities inspection and preventive-maintenance platform. It brings equipment, recurring tasks, findings, work orders, and records into one place.

Our planned AI work focuses on helping users set up their site and migrate records from existing maintenance systems. A setup assistant could build an asset model from a conversation; a migration assistant could map imported records and propose schedules for human review.

CUDA Toolkit, NVIDIA NIM for Llama 3 8B Instruct, and NVIDIA NeMo Agent Toolkit are under consideration. These are planned capabilities, not claims of NVIDIA acceleration in the current service.

Visit FORC Inspection →

NVIDIA Inception

Part of the startup community.

NVIDIA Inception gives startups access to developer resources, training, and opportunities across NVIDIA's ecosystem. Our membership sits alongside hands-on work on local AI and embedded systems.

Membership identifies our participation in the program. The product descriptions and development plans on this page are Oregon Radio Company's own; membership does not imply NVIDIA endorsement or certification of these products.

About the NVIDIA Inception program →

© 2025 NVIDIA, the NVIDIA logo, CUDA, JetPack, NVIDIA Inception, NVIDIA NeMo, NVIDIA NIM, and TensorRT are trademarks and/or registered trademarks of NVIDIA Corporation in the U.S. and other countries.