Overview
Vigilo was developed as a security and moderation system for online communities. The project explores how machine learning and automation can assist with identifying scams, phishing attempts, and potentially malicious user content.
← Back to Projects
CYBERSECURITY · MACHINE LEARNING
Vigilo
A security-focused Discord moderation system combining machine-learning classification, OCR analysis, access controls, automated enforcement, and audit logging.
922K+ samples
0.94 ROC-AUC
0.93 Average Precision
Machine Learning
The classification pipeline was trained and evaluated using more than 922,000 labeled samples. OCR-extracted text can be passed through the classification system to support detection of suspicious content contained inside images.
Evaluation produced an average precision of approximately 0.93 and ROC-AUC of approximately 0.94.
Security & Automation
Vigilo includes automated quarantine and enforcement workflows, role-based access controls, audit logging, appeals, configurable detection thresholds, and mechanisms for managing false positives.
What I Learned
Building Vigilo has allowed me to combine software engineering, machine learning, access control, security automation, API integration, and practical moderation workflows within a single system.