• Direct CEO report; member of executive leadership team. Budget ownership for technology spend, strategic planning participant, vendor management, end-to-end RFP and pricing strategy for deals ranging from 50 to 100,000 users.
• Product roadmap gatekeeper — every new feature requires technical blessing. Hired every technical staff member for the past 10 years; 5+ year average engineering tenure.
• Currently positioned as CTO ahead of potential company acquisition; expected to lead due diligence and architecture presentations.

• Architected the entire Generative AI service layer powering all AI features in the platform: agent framework, RAG system, streaming infrastructure, and responsible-AI / quality layers.
• Built RAG system with Amazon Titan embeddings, S3 Vectors, hybrid chunking (size + structure), and semantic search across diverse content types. Designed multi-judge validation pipeline (Anthropic Claude + Amazon Nova) with automatic regeneration on quality failures.
• AI Assessment Generator: reduced generation time from minutes to single-digit seconds via topic-grouped embeddings — accuracy increased simultaneously because models weren’t drowning in irrelevant material.
• Designed for AI governance (ISO/IEC 42001 alignment): per-tenant feature flags, full data isolation, no cross-customer data leakage, no new attack surface vs. existing SOC 2 posture.

• Own the entire multi-tenant AWS architecture serving 80+ production clients with per-tenant isolation at every layer: 383 S3 buckets, 161 CloudFront distributions, 133 SQS queues, 219 CloudFormation stacks, ~175 Secrets Manager entries, 38 ECR repos, 28 Route53 zones.
• 6 EKS clusters across us-east-1 and ca-central-1 (production, sandbox, dev, internal, CA-production, CA-sandbox) for geographic data residency.
• Built ADEPT internal cloud deployment management platform from scratch: CloudFormation orchestration, Kubernetes admin, real-time WebSocket status, fully-automated PR preview environments mirroring production, AI-powered RCA chatbot for Customer Success.
• Replaced Rustici Engine SCORM runtime with in-house solution backed by 6,000+ conformance tests (50% vendor cost reduction). CloudFront caching layer reduced API latency by ~95% and external API costs by ~80%.
• 99.9–99.99% uptime SLA with self-healing infrastructure — systems auto-recover within 1–2 minutes, eliminating on-call rotations. 1-minute PITR via Aurora.

• Led technology controls for first SOC 2 Type II certification (Q1 2026): owned majority of audit controls (infrastructure, access management, deployment, monitoring, data security). Auditors noted exceptional readiness for company size.
• Architected enterprise identity infrastructure on Keycloak 26.x (7-repo ecosystem): per-tenant realm architecture with OIDC/SAML SSO, custom electronic-signature authenticator for 21 CFR Part 11 compliance, custom themes, self-service IdP management UI.
• Automated compliance documentation tooling generates 600+ page audit-ready validation package (IQ/OQ/PQ) auto-linked to test execution.

• Designed and built Ratchet, an enterprise-grade distributed job scheduler for Jakarta EE (now being open-sourced): cluster-safe via FOR UPDATE SKIP LOCKED, pull-based architecture, adaptive polling reducing DB queries up to 95%, type-isolated thread pools, virtual thread support, built-in circuit breakers. Replacing 66 legacy jobs.
• Maintain scorm-again (303+ stars), mjml-java, elearning-module-parser, and agent-peer-review (28+ stars).

• Combined management, system administration, client interaction, and hands-on development. Spent 25–35% of time writing and developing features.
• Migrated entire infrastructure from on-prem datacenter to AWS, reducing monthly hosting costs by 75%+. Transitioned database from Oracle to Amazon Aurora MySQL.
• Migrated deployments from manual upgrades to containerized (Docker) instances on Kubernetes clusters; established GitOps for application deployment and automated upgrades.
• Led complete rewrite of the assignment logic layer (LMS core business logic for content assignment, due dates, completion rules, training plans).
• Led UI modernization: migrated 200+ pages from JSP to Jakarta MVC (Krazo) with Handlebars.java templates; maintained handlebars-java fork.
• Created and implemented IT security policies. Handled all code reviews and merge requests.

• Returned to lead development modernization. Upgraded core Java/JBoss application to current versions; migrated SVN to Git; set up Jenkins CI.
• Modernized application UI from table-based layouts to responsive design (Bootstrap, HTML5, CSS3); migrated Flash assessment module to HTML5 + jQuery.

• Coordinated Kenexa 2x BrassRing system testing for the Progress Energy / Duke Energy merger; supported PeopleSoft HRMS with SOX-compliant access controls.
• Developed SharePoint-based automated reporting dashboard (jQuery + Google Charts API), reducing weekly/monthly report development time by 80%+.
• Automated data file encryption and transmission with Perl. Managed Sarbanes-Oxley compliance initiatives.

• Entry-level role on the LMS. Java, JSP, ActionScript, MXML (Adobe Flex).

• Modern JavaScript runtime library for SCORM 1.2 and SCORM 2004. v3.0 (Dec 2025) shipped a full SCORM 2004 sequencing engine (SN Book specs), 95%+ test coverage, 6,000+ conformance tests, Playwright cross-browser tests, and a complete Docusaurus documentation site.

• Pure-Java MJML 4 email renderer with zero runtime dependencies beyond the JDK. All 31 MJML components, 1,116 tests including 40 golden-file tests matching the official MJML CLI output, Spring Boot auto-configuration module.

• CDI-native distributed job scheduler for Jakarta EE 10. Cluster-safe via FOR UPDATE SKIP LOCKED, no Redis/Zookeeper required. Multi-module structure with TCK for validating custom store implementations. 50+ Arquillian integration tests on WildFly.

• Parses every major eLearning standard: SCORM 1.2, SCORM 2004, AICC, cmi5, xAPI/TinCan. Rule-based validation architecture with 30+ composable validation rules, security-hardened XML parsing, multiple file access backends (local, S3, in-memory, ZIP).

• Claude Code plugin marketplace for AI-to-AI peer validation. Two AI perspectives catch more issues than one — uses OpenAI Codex CLI as a peer reviewer with a structured 2-round discussion protocol for disagreements.