Engineering Depth.Executive Trust.

Problem

Document retrieval took analysts 15+ minutes per lookup, and the client had never put an AI product in front of its users.

Approach

I served as delivery owner and lead tech strategist for a greenfield RAG application on AWS Bedrock and OpenSearch. I defined the technical approach, refined retrieval architecture, and integration approach across data engineering, infrastructure, security, and application teams, then ran 9 production deployments in 6 months while iterating on user feedback.

Result

0 to 16,000 users in 6 months. Research cycle times down 93%, an estimated 1,000+ hours saved a year. The work earned a follow-on contract to expand into new use cases and business units, with a roadmap to 40,000 users.

Problem

A ServiceNow environment on AWS GovCloud runs the Employee Portal, HR processing, and new hire onboarding for 65,000+ users and 300,000+ HR transactions a year. It has to keep shipping without disrupting any of them.

Approach

I owned implementation and ongoing platform health: APIs, integrations, and infrastructure dependencies coordinated through 2-week sprints. I redesigned intake, prioritization, and deployment practices, and ran proactive health monitoring that surfaced and resolved 20+ portfolio risks a year through early escalation.

Result

An HR platform built around ServiceNow HRSD: a custom service portal, batch record initiation, SSO sign-in, source record and data integration, and microservices for capabilities ServiceNow does not ship natively. Over seven years, it carried 150+ releases with fewer than 10 critical defects, and automated new hire onboarding components replaced paper processes entirely for $90,000 a year in documented savings.