This study presents a hybrid computational framework that integrates structural analysis, dimensionality reduction, and metaheuristic algorithms to optimize the performance and design of steel structures. Leveraging ETABS, a widely used structural analysis software, the framework automates the extraction of key responses under various load combinations. To address complexity and high dimensionality of structural data, dimensionality reduction techniques, including Principal Component Analysis (PCA) and t-distributed Stochastic Neighbor Embedding (t-SNE), are employed. The dimensionality reduction method involves training an AutoEncoder and a deep feedforward network to compress the high-dimensional training samples into a low-dimensional latent space. To validate the quality of the reduced space, supervised machine learning classifiers such as XGBoost, LightGBM, and CatBoost are trained to distinguish between feasible and infeasible solutions in the transformed domain. These classifiers evaluate label separability after dimensionality reduction, without influencing the optimization process itself. Metaheuristic algorithms, such as Particle Swarm Optimization (PSO) and Symbiotic Organisms Search (SOS), are implemented to explore the vast design space and identify optimal solutions. These algorithms are tailored to minimize structural weight, maximize safety, and ensure compliance with design codes while reducing computational costs. The proposed framework seamlessly integrates these tools, enhancing the decision-making process for engineers and designers. Computational performance is benchmarked using benchmark problems and a high-rise 20-story braced steel frame. The results demonstrate that the proposed framework significantly reduces computational cost and dimensionality without compromising solution quality. The reduced space enables more efficient optimization, while the classifier-based evaluation confirms the structural fidelity of the transformed feature space. This research contributes to advancing computational optimization in structural engineering, offering a flexible tool for designing diverse structural systems. Future work includes integrating machine learning models for predictive analysis and expanding the framework to other structural materials and systems.