Work
Six years of platform, cloud, and developer-experience work at John Deere, and a current
chapter spent making AI a first-class participant in how we ship software. Most of it sits
behind the enterprise firewall, so these case studies focus on architecture, constraints, and
outcomes rather than source links.
Where my work points now: treating AI as a first-class participant in the software development lifecycle. That means the harness that decides whether agent output is safe to ship, the context layer that makes it useful, and the measurement that replaces “AI helped” with evidence.
2026 – present
vibes → evidence · Measuring AI’s real impact
On a three-person tiger team chartered by senior leadership, I’m defining how agentic AI reshapes the software development lifecycle, then building the harness, context substrate, and workflows that turn those findings into practice. I architect the infrastructure that makes AI agents safe, trustworthy, and effective: context platforms, orchestrators that plan and verify, and guardrails for security, observability, and compliance. The throughline is replacing ‘AI helps’ vibes with objective evidence, and operationalizing the finding that context, not the model, is the load-bearing layer of agentic delivery.
claude-code codex copilot mcp github-actions knowledge-graph
Read case study →
2025 – 2026
On the developer-experience team that owns the path from pull request to production, I built the tooling, governance, and automation that let the enterprise ship fast with security and separation-of-duties built in. The work spanned an event-driven compliance platform monitoring 50,000+ repositories, enterprise incident response, and AI-assisted diagnostics that cut production investigation from hours to minutes.
aws github-actions event-driven python typescript splunk
Read case study →
The platform years the AI thesis is built on: compliance automation across 50,000+ repositories, a network consolidation that saved seven figures, and the unglamorous plumbing that let other teams move fast.
2023 – 2025
Led an enterprise-wide network consolidation across ~2,000 VPCs, optimizing NAT Gateway usage to cut annual cloud spend by $1M. Orchestrated a DNS migration that remediated 140,000+ security vulnerabilities with 100% uptime during cutover, and migrated 10,000+ AWS ACM certificates from email to automated DNS validation, making renewals fully automated enterprise-wide.
aws vpc transit-gateway route53 terraform acm
Read case study →
2023 – 2025
I owned AWS account automation and the operational plumbing thousands of engineers depended on daily. Replaced a manual, hour-long account and VPC deletion process with an event-driven Step Functions and Lambda system (hours to seconds), designed out DNS subdomain-hijacking risk across ~900 hosted zones, and cut developer onboarding from hours to minutes with Dev Containers.
step-functions lambda event-driven python route53 dev-containers
Read case study →
2021 – 2023
On the License Management team, behind a high-volume customer-facing web application accounting for roughly 10% of John Deere’s annual revenue, I built a secure data-ingestion pipeline during a hackathon that proved the case for making license data available to internal teams while preserving isolation and access controls. The prototype directly contributed to forming the enterprise Data Engineering team.
java aws data-pipeline oauth
Read case study →
2019 – 2021
On the Global Gateway API platform, I architected the platform's first external-facing API (designed from the start with role-based authorization and audit logging) so dealers could safely access sales-lead data at John Deere Financial.
java api-gateway oauth rbac
Read case study →