Processed and triaged 200+ geolocation data defects per week during the Apple Maps global launch. Partnered directly with GIS engineers to improve regional map accuracy across three metro areas, and helped establish the geo-data validation pipelines and QA automation frameworks used during rollout.
Led full migration from a legacy codebase to a modern MVC web architecture. Significantly improved maintainability, performance, and readiness for the AI and analytics features that followed over the next decade.
Architected ad tracking infrastructure integrated with a proprietary CRM. Processed over 1.5M ad visitor sessions and $2.5M in ad spend, enabling revenue attribution across Google, Meta, and email. Correlation analysis identified high-ROAS channels, informed LTV strategies, and shaped campaign targeting.
Built an ASP.NET photo management system to store, render, and display customer-submitted images tied to historical orders. Metadata tagging and classification features later enabled AI-powered visual search and retrieval. Integrated with legacy SQL Server and proprietary email processing.
Built a proprietary ASP.NET engine that displays real-time color renderings of embroidered items. Integrated with CRM and fulfillment systems and structured to support future ML-based personalization and product preview enhancements.
Designed an ASP.NET system to automatically solicit customer reviews via email and SMS, with messages personalized using product images, order metadata, and delivery timing. Incorporated basic NLP sentiment analysis to flag negative responses and trigger support follow-ups — laying the foundation for AI-driven customer satisfaction loops.
Reengineered internal security architecture via FortiGate firewall deployment and workstation-level monitoring. Improved network resilience and compliance readiness, positioning the company's infrastructure for AI system data protection requirements.
Modernized payment infrastructure to meet PCI DSS compliance standards. Laid groundwork for AI fraud-detection integration as a future layer.
Spearheaded federal grant acquisition for PPP and ERTC programs by automating documentation workflows and streamlining internal financial reporting. Executed end-to-end under extreme operational constraints during the pandemic.
Led R&D and full delivery of a new product category integrating software enhancements, CRM/logistics updates, and CO2 laser hardware setup. Demonstrated full-lifecycle ownership from manufacturing integration to digital pipeline deployment.
Designed and deployed server-side tracking infrastructure using AWS API Gateway with proxy-based data handling for GA4 and Meta. Enabled GDPR-compliant analytics protected against browser-side limitations, establishing a high-fidelity cloud data stream for predictive modeling and AI-driven customer segmentation.
Consolidated legacy mobile views into a responsive web architecture. Created a unified front-end foundation ready for ML-driven personalization layers.
Led remediation of critical network and browser security vulnerabilities uncovered in a third-party audit. Aligned infrastructure with current best practices for AI model data protection.
Modernized a 300K+ LOC legacy .NET 3.0 monolith into containerized microservices deployed on Azure. Enabled full integration with AI agents across chat, voice, text, and email — all connected to legacy CRM, ERP, and logistics. Used self-hosted n8n as the orchestration layer for event-driven AI automation.
Designed and built AI support agents capable of handling customer inquiries via chat, email, text, and voice — entirely from scratch. Integrated OpenAI LLMs with a legacy Microsoft SQL Server backend using n8n workflows and contextual query guardrails, enabling natural language access to structured business data. Incorporated real-time speech-to-text for voice interactions. The chat agent running on this page is the same system, connected live to a real backend.
Conversational AI workflow with persistent session memory stored in Postgres. Full transcript logging enables quality review, audit trails, and retraining data capture from real interactions.
Most of this work is proprietary.
The architecture isn't.
I'm happy to walk through system design, decisions made, and lessons learned for any project on this page — no code required.
Request a walkthrough