Taehun Choi

This project modernized network equipment monitoring by replacing static threshold-based alerts with an AI-driven anomaly detection system. Instead of configuring device-specific thresholds, the solution learned each device’s normal traffic patterns and automatically triggered alerts via email and messenger when abnormal behavior was detected.

To address varying usage patterns by time of day and weekday/weekend, the system applied time-series machine learning models to classify device states as normal or abnormal. It also incorporated predictive analytics to forecast potential failures, enabling both real-time anomaly detection and predictive maintenance.

As an end-to-end contributor, I led architecture design, development, and deployment. I also implemented large-scale topology rendering for over 2,000 network devices, using D3.js to deliver interactive, real-time visualization that improved situational awareness for operators.

Backend Architecture

Frontend Architecture

Software Stack