Industrial AI from strategy through PLC — built by the same engineer who designs it.

When a manufacturer asks me to bring AI to a line, the engagement usually starts with AI Consulting — vendor-neutral architecture before code. From there it might branch into a PLC retrofit if the cabinet needs modernizing, or a system prototype if the integration shape isn't yet defined. When the orchestration layer matters, NexusVisionOrchestrator is the platform I bring; when the vision model itself is the bottleneck, CodingNexusVision is the engine I extend.

Three principles run through all of it. Vendor-neutral by default — the orchestrator drives Maddox Edge, Estek, CodingNexusVision, or any ONNX model in the same recipe, so the architectural decision survives a vendor change. Line walk before code — I refuse to design a PLC retrofit from a Zoom call; the first invoice always covers a physical or video-attended cabinet survey. Real data in the pilot — every pilot runs on production frames the customer captured, not on the vendor's curated demo dataset, because that's where the brittleness lives.