At FlyRank AI, I built a Python pipeline to analyze 30,000+ anonymized search-performance records and flag organic traffic declines and content refresh opportunities. I trained a Random Forest model with cross-domain holdout validation, reaching 60% Precision@50 and outperforming the traditional SEO heuristic baseline by 12 percentage points.
I also built AppMonitor, a full-stack application monitoring system with endpoint health and API latency checks, severity-based alerts, and deduplication. For PhishSafe, I developed a phishing email detection project that achieved 96.17% test accuracy and a 0.95 phishing F1-score, with real-time on-device JavaScript inference.
