An SQL injection vulnerability can be either an unintentional mistake by developers or an intentional attempt by hackers to exploit sensitive information. With the increase in today's information overload, there is a natural urge to protect this data from going into the wrong hands, which could lead to theft, leakage of confidential information or some misleading damages. Relational databases such as MySQL are the most famous, allowing its users to draw out all accessible data without deep and significant database knowledge. Because databases store so much information, making an attractive information target or attention of attackers who can compromise potentially sensitive and critically confidential information. Early detection of SQL injection attacks can go a long way in preventing malicious attempts by attackers. In this study, we analyse the performance of various supervised & Deep Learning algorithms such as Random Forest, Decision Tree, CNN, RNN, SVM, etc. with reinforcement learning algorithms such as Q-learning, by combining the datasets consisting of potential SQL injection queries. We aspire to deliver a cutting-edge reinforcement learning solution aimed at ameliorating the looming spectre of SQL injection. Employing state-of-the-art machine learning algorithms, our model is designed to adeptly discern SQL attacks, accelerating its capacity to preempt unforeseen anomalies with heightened precision and celerity.