CodingNexusVision — an industrial vision-AI engine with a clean API and no vendor lock-in.

CodingNexusVision is a focused industrial vision engine — it ingests images and returns verdicts (classification, defect localization, score). It does not try to be a full orchestration layer; for that, it is consumed by NexusVisionOrchestrator or any other system that speaks its API. This separation is deliberate. Most vendor offerings bundle an engine, an orchestration layer, an annotation tool and a deployment portal into one product, then license them as one product. Keeping the engine separate means you can swap orchestration without swapping the engine — and vice versa.

Three properties make it usable in industrial deployments. Deterministic resource usage: GPU memory footprint, CPU time per inference and latency variance are characterized at startup, not discovered in production. On-premise by default: no outbound traffic, no telemetry, offline activation. API-first: REST for control-plane operations, gRPC for high-rate inference, no SDK lock-in.