A structured multi‑agent LLM framework with a JSON‑serializable Design‑State Graph for early‑stage systems engineering.
Unified framework integrating Panel/Bokeh GUI, ensemble neural networks, and ParaturboCAD for real-time simulation and automated CAD generation of gas-bearing supported turbocompressors.
Automated 3D CAD generation of entire turbocompressor spindles from a single Python-scripted workflow, reducing manual geometry handling to 7 minutes per design.
Automated, Robust, Modular and Integrated Design framework - the backbone of the entire thesis enabling automated optimization of complete turbocompressor systems under constraints.
