Given the fast expansion of digital infrastructure, DoS and brute-force attacks raise serious issues. These risks compromise data security, disturb system resilience, and influence service availability. This study proposes a multi-layered defensive solution integrating machine learning, encryption technologies, and anomaly detection to lower the impact of DoS and Brute Force attacks using an improved cybersecurity framework. The proposed architecture detects aberrant traffic patterns suggestive of DoS assaults using supervised learning models with an intelligent intrusion detection system. Further actions to thwart volumetric DoS attacks rooted in protocols include rate-limiting techniques and CAPTCHA authentication. We provide an adaptive authentication method grounded on user behaviour to lessen the impact of brute-force attacks. This method prevents unauthorized access attempts using dynamic account lockout rules, MFA, and encryption-based password hashings. To top it all, we have integrated real-time threat data to increase predictive security, speed response times, and the detection accuracy of shifting attack patterns. Benchmark datasets reveal that false positives have decreased while attack detection rates have significantly increased. Comparative findings reveal that using more conventional methods, the proposed methodology is superior in reducing system vulnerabilities and ensuring high security. This research promotes cybersecurity by offering a proactive and intelligent defence system against DoS and Brute Force attacks. Studies indicate that solutions powered by artificial intelligence are needed to protect digital infrastructure against new cyber threats.