Kuzushi — Agentic SAST scanner with AI triage

Official repo of "EndoBoost: a plug-and-play module for false positive suppression during computer-aided polyp detection in real-world colonoscopy (with dataset)"

GUARDIUM is an intelligent Wazuh rule optimization framework designed to reduce false positives, improve alert accuracy, and assist SOC teams in maintaining high-quality SIEM detections. GUARDIUM combines rule analysis, threat context, and Large Language Models (LLMs) to automatically evaluate, explain, and optimize Wazuh rules.

Analytical portfolio demonstrating transaction monitoring, judgment-based alert review, and Excel-driven risk analysis across fraud, AML, and KYC workflows, with a focus on regulator-safe decisioning and operational consistency.

🚀 SmartAlert AI - Production-ready adaptive ML system for intelligent incident prediction. Achieves 44% F1-Score on challenging datasets with 75% false alarm reduction. Features 118 sophisticated features and real-time adaptive learning.

Python security gate with intelligent ML scoring that reduces false positives by 95%. Orchestrates Bandit, pip-audit, and Semgrep into a unified CI/CD pipeline. Includes baseline management, policy enforcement, and explainable predictions. Production-ready with comprehensive tests.

Lightweight network based host & service watchdog - very smart check definitions to cut down on false positives. Eight different nest-able, AND/OR-able health check types. Email, Signal and Webhook alerts. SOC/NOC wallboard, SLA tracking and more.

Research-grade architecture for memory-augmented, agentic AI in SOC alert analysis. Combines Episodic, Semantic, Procedural, and Working memory with LLM reasoning (FastAPI) for context-aware threat analysis and false positive reduction. Prototype; not production-validated.

A Claude Skill that adjudicates sanctions / PEP / adverse-media screening hits deterministically. Token-efficient tiered design (parse → cheap FP rules → identifier corroboration → targeted multilingual research). Handles 9+ naming conventions, transliterations, and historical place names.

🔍 Analyze transaction data to identify fraud risks, streamline alert reviews, and ensure compliance in AML and KYC contexts using Excel-driven techniques.