AI Engineer Achieves Certifiable Physical Engineering Design with Closed-Loop System

While AI has made strides in scientific discovery, complex physical engineering design, which requires satisfying multiple simultaneous constraints across disciplines like fluid dynamics and structural stability, has remained a challenge. A new agentic framework, "The AI Engineer," addresses this by integrating large language models (LLMs) with deterministic engineering simulation backends in a closed-loop system. This framework begins by converting natural language requirements into design geometry. It then uses bi-directional evolutionary structural optimization (BESO) with the CalculiX solver for topology optimization, followed by particle swarm optimization (PSO) with Zwind for refining member sizes under specific load cases for offshore floating-wind projects. To efficiently explore numerous designs, an Automated Reviewer scores each candidate on five dimensions (capacity, steel intensity, unit cost, constructability, and fatigue life) using calibrated functions. The design process terminates only when a candidate achieves a composite score of 85 or higher, with no subscore below 60. This rigorous gate was validated by submitting the top-scoring design to the China Classification Society (CCS) for Approval in Principle (AIP), which it successfully passed. This external certification confirms the reviewer's alignment with professional judgment. The AI-certified design surpassed a human-optimized baseline, reducing steel mass and unit capital cost by 8.1% each while meeting all AIP criteria. This verification-closed approach, where every proposal is judged by deterministic physics and codified limits, distinguishes The AI Engineer from open-ended generative systems.