In the modern world, the prevalence of Autism Spectrum Disorder (ASD) is increasing, affecting approximately 1 in 54 children—a challenging neurodevelopmental condition. The early and precise identification of ASD is crucial for timely interventions and support for individuals and their families. This research paper delves into the domain of machine learning, aiming to harness its potential in enhancing ASD detection. To achieve this, a comprehensive dataset encompassing clinical, behavioral, and genetic information from both ASD and non-ASD individuals is leveraged. Various machine learning algorithms, including Decision Trees, Support Vector Machines, and Deep Neural Networks, are applied. Additionally, techniques like feature selection and engineering is integrated to enhance model performance and interpretability. The study yields promising results, demonstrating the effectiveness of machine learning in distinguishing between ASD and non-ASD cases. The research emphasizes the importance of model interpretability, providing insights into the key features influencing ASD classification. This data-driven approach empowers clinicians to make well-informed decisions, enabling early diagnosis and intervention for individuals with ASD, ultimately contributing to improved outcomes and quality of life. The newly devised Hybrid CNN model which is a combination of CNN with SVM outperformed all other basic models and gave 96.8% precision and 93.01% accuracy in determining autism disorder in children.