From threat analysis to production AI.

My career in applied AI began with cybersecurity work. At Cisco Talos, I spent years analyzing large-scale security data, maintaining threat-intelligence databases, automating reporting, and developing approaches to improve phishing URL detection. The work made one pattern hard to ignore: analysts repeatedly faced classification problems that machine learning could help structure and scale.

That observation became a transition into data science and machine learning—not away from security, but deeper into it. I learned to treat features, labels, and evaluation as parts of an operational decision system. A model was useful only if it could survive changing adversary behavior, incomplete context, and the consequences of false positives.

At Cisco Learning & Development, I expanded that systems perspective into generative AI. I led a services team delivering a production retrieval-augmented generation assistant on AWS, built semantic search and knowledge-retrieval experiences, and used telemetry and feedback to improve the product. Related work in graph data models and metadata reinforced the same lesson: context and provenance shape quality.

Today, as a Senior Data Scientist at ThreatSTOP, I lead machine learning pipeline work for phishing detection in DNS traffic. The system combines character-level language features and multiple classifiers with WHOIS, DNS, VirusTotal, DGA, and anomaly signals. It is designed around real constraints such as shared API quotas, deduplication, model artifacts, and security-team workflows.

I want to keep specializing at the intersection of AI security and production applied AI—work where model quality, system reliability, and technical judgment all matter.