AI-driven engineering, from requirements to part

The situation

Engineering tools were built for people, not for AI, and the legacy systems didn't talk to each other. The question was whether AI could carry a part from customer requirements all the way to a machine-ready file on one connected digital thread.

What I did

• Worked with the solutions architect to design the end-to-end solution, and was hands-on at every stage.
• Built LLM workflows that turned customer needs into requirements written to international standards and modeled in Cameo (MBSE).
• Built the data converters, LLM integrations and Istari Digital integrations that joined the stages, and helped onboard Istari Digital with their forward-deployed engineers.
• Piped AI-generated CAD (Zoo) into meshing, CFD and FEA, then into VR/AR for review, then out to a machine-readable manufacturing step.
• Consulted the subject-matter experts on what each stage needed, and coordinated the teams so they actually used the pipeline.
• Earlier, built a CAD → OpenUSD → NVIDIA Omniverse pipeline for real-time design review, and connected SteamVR to Omniverse.

What changed

• A working path from requirements to a machine-ready part on one digital thread, delivered in about six months.
• CEO-level visibility unblocked people's time and the backend work that the legacy systems had been holding up.