The increasing adoption of IoT technologies has led to a heightened emphasis on network security. However, due to limitations in resources and the intricate nature of algorithms, conventional ML intrusion detection methods face challenges when implemented within IoT networks. The leverage of the advancements of swarm algorithms has proven their efficiency in solving complex problems. This study aims to design, implement, and test an IDS placement strategy based on traditional ML algorithms with hyperparameter tuning. A population-based swarm algorithm called Firefly is introduced for extracting the pertinent parameters of ML algorithms through hyperparameter tuning. This framework demonstrates effective performance when applied to the TON_IoT dataset featuring new variables. Additionally, an analysis comparing the predictive outcomes of the proposed hyperparameter tuning method against those obtained using default parameters is conducted. The outcomes of the experiment validated that the suggested IDS framework designed for IoT successfully identifies genuine attacks, thereby bolstering the security within the IoT ecosystem. Additionally, fine-tuning the hyperparameters resulted in striking a balance between efficacy and resource utilization, leading to superior accuracy when contrasted with utilizing default ML parameters.