Published Paper Details:

The **Smart Intrusion Detection System (Smart IDS)** is a hybrid AI/ML framework designed to secure cloud environments against sophisticated threats, including brute-force attacks, malicious query execution, and unauthorized data extraction. Recognizing the limitations of conventional systems--specifically the poor adaptability of static rules and the high false-alert rates of isolated machine learning models--this system integrates **deterministic threat heuristics** with **Isolation Forest-based anomaly detection**. A **weighted risk scoring model** ($R = 0.7 cdot S_{r} + 0.3 cdot S_{m}$) dynamically categorizes incidents into operational severity tiers. To alleviate analyst workload while maintaining data privacy, the framework utilizes a **locally deployed Large Language Model (Mistral)** to generate incident summaries and mitigation guidance for high-severity alerts, ensuring sensitive logs are not shared with external parties. The platform features a **real-time React-based dashboard** for monitoring risk trends, geographic attack sources, and active alerts. Experimental results indicate that the hybrid approach achieves **93.4% accuracy**, significantly outperforming standalone rule-based or machine learning models.