Internal AI agent platform at a Fortune 50 technology company.
Built and continue to operate the framework that hundreds of engineers at the company use as the standard starting point for new internal AI agents. The platform handles authentication, model abstraction, safety guardrails, security scanning, telemetry, and orchestration as a single layer beneath each new agent.
The platform is updated weekly. Engineering teams that previously took six to eight weeks to deploy a new internal agent now deploy in days, with the security and safety controls already in place. The platform is not productized externally. It is internal infrastructure, treated as such.
Nation-state threat defense in a classified federal environment.
Led security operations in an air-gapped federal facility against advanced persistent threats, including spearphishing campaigns and intrusions using professional command-and-control frameworks. The environment was physically secured, air-gapped from the public internet, and subject to the operational constraints of classified work.
The engagement spanned multiple years. Specific tradecraft, indicators, and attribution remain classified and are not discussed in any forum.
AI model security scanner for a hyperscale machine learning pipeline.
Designed and built an automated vulnerability detection system that runs against models inside a hyperscale ML pipeline. The scanner performs model lineage tracking, dependency analysis, and AI-specific attack path enumeration. It is integrated into the pipeline as a required gate, with coverage exceeding ninety-five percent of production models.
The scanner runs without operator intervention and produces machine-readable findings that route directly into the existing security incident workflow. False positive rates are tracked and tuned against red-team output.
Autonomous security analyst for small and mid-market organizations.
Architected and delivered a production autonomous AI agent that connects to a customer's existing security tooling, including SIEM, EDR, and identity systems. The agent investigates alerts twenty-four hours a day. It operates within explicit autonomy tiers configured per tenant, with administrative kill switches, per-tenant isolation, and tamper-evident audit logging.
The system runs in production. Multi-tier autonomy controls and prompt-injection defenses are documented in the delivery doctrine.
Federal contracting automation for government contractors.
Built an autonomous agent system that discovers federal contract opportunities, ingests solicitation documents, generates compliance-mapped proposal drafts, and replaces the labor traditionally performed by capture managers and proposal writers. The system operates against SAM.gov opportunity data, contracting officer documentation, and customer-supplied past performance records.
Quality gates run between every stage. Drafted proposals are subject to compliance checking, evaluator-criteria alignment scoring, and finish-plan generation before any human review.