<p>The potential of indoor localization to revolutionize location-based services within enclosed spaces is remarkable. Navigating in Internet of Things (IoT) based environments has become a focal point in various fields such as healthcare, security, and tracking. This study begins by reviewing state-of-the-art research on indoor localization and provides a comparative analysis of these works. Subsequently, the study employs different Wi-Fi fingerprinting algorithms on a proprietary dataset. Various machine learning techniques, including KNN, RF, and ANN, are applied, coupled with optimization algorithms such as Gradient Descent (GD) and Grid Search (GS), to estimate the latitude and longitude of mobile devices. GS is utilized for the parameter optimization of the machine learning algorithms. The results include the identification of the building, floor, and positional error assessment. Our most accurate model, ANN-GD, achieved 100% accuracy in building identification and 90.5% accuracy in floor identification and overall accuracy of 97.9%. Proposed model surpasses the benchmark by approximately 5%. These results show the potential of optimized machine learning models for real-world indoor positioning systems to facilitate better navigation, emergency response and smart infrastructure development.</p>