Yusuf Morsi

Owned and led the link analysis automation project end-to-end, cutting a manual two-day data processing pipeline down to a few hours, with results surfaced across heatmaps, pinmaps, a RAG chatbot, and more- all built from scratch and deployed as an internal dashboard at link-analysis.cisco.com.

Built optimal RAG systems with attribute filtering from scratch for device log diagnosis.

Developed a multivariate anomaly detection process analyzing thermal telemetry across tens of thousands of production network devices to diagnose future corrosion beforehand purely from data.

Benchmarked six anomaly detection models (including CADE, Isolation Forest, and autoencoders) on fraud transaction data from major banks across the US, UK, Canada, India, and Brazil.

Engineered rolling temporal features that improved AUC and precision during distribution shift periods. Presented findings directly to the Chief Analytics Officer.

Led development of a license plate recognition pipeline using YOLOv4, TensorFlow, and OCR. Built and published a pip package for the company's API alongside full-stack work in Vue, Nuxt, and Next.js.