Multi-Node Threat Intelligence Pipeline (Honeypot)
Wanted one project touching everything learned: cloud, automation, DevOps, AI integration, data, visualization, security. Not another off-the-shelf Cowrie/T-Pot deployment.
Built and operated a 3-node CTI pipeline on DigitalOcean across 3 months, capturing, deduplicating, and openly publishing a 47.6K-log threat dataset (Zenodo DOI) and reporting 1,755 high-severity threat IPs to AbuseIPDB. Engineered an automated GitHub Actions backend using Vertex AI Gemini API for payload classification; audited LLM severity labels against human review (n = 100), improving inter-rater reliability from Cohen’s κ = 0.66 to 0.80 by correcting systematic labeling inconsistencies. Enforced Principle of Least Privilege via Linux kernel capabilities (cap try via an interactive 3D threat globe (Three.js/Tailwind CSS).
47,598 total probes, 5,095 unique IPs, 11,366 critical and 14,501 high severity classified, published openly as a dataset with a Zenodo DOI, and 1,755 high-severity IPs reported to AbuseIPDB. Audited AI severity labels against 100 human-reviewed samples, improving Cohen's kappa from 0.66 to 0.80.