I’m Mitch. I build platforms that make reliable software delivery easier.
Recent work · October 2025 – October 2026
The AI-Era SDLC
I’m a Cloud Software Engineer at John Deere, working across developer platforms, cloud infrastructure, and AI-assisted delivery. On a three-person team, I connect product definition, backend implementation, and the controls needed to move a system into production. Read my experience or download my résumé →
Production delivery
For a team-owned engineering evidence product, I wrote the requirements and much of the design record, delivered key live-data views and a personal work timeline, and designed and implemented shared caching. The product is in production with steady, growing use. The caching decision balances latency, data freshness, and availability; the case study walks through that tradeoff and its acceptance checks.
Cloud execution and trust
I ran Codex and Claude Code in cloud sandboxes and added repository-credential proxying so the credential stays outside the execution environment. I also built a skill-scanning trust layer and packaged reusable review tools. A separate keyless Codex pilot has merged; live verification and containment hardening remain in progress.
Operating the delivery platform
In late 2025 and early 2026, I hardened the separation-of-duties platform with Datadog monitors, dashboards, and synthetics, added automatic incidents for pipeline failures, and consolidated a legacy processor. I also worked through security findings and reviewed organization-governance changes.
Making the work reusable
I designed the shared team brain and its retrieval, graph, memory, and quality tooling. I wrote the team’s harness thesis and Agent Delegation Stack, and co-facilitate a guild and office hours for around 38 practitioners. Sourced briefs, repository instructions, review skills, and feedback hooks make that work available beyond an individual session.
My working model is Agent = Model + Harness: model capability becomes useful through context, controls, tests, and feedback. I explore it in the Agent Harness field guide. Production adoption and observational signals do not establish causal productivity gains.
Earlier foundations · 2019 – 2025
From APIs to Enterprise Platforms
2019 – 2021 · John Deere Financial
I started at John Deere Financial during an early attempt to make APIs safely available to dealer software. The ambition resembled what a product such as Azure API Management provides: a common gateway for governing and exposing APIs instead of another one-off integration. I helped architect the platform’s first external-facing API for dealer access to sales-lead data, with role-based authorization and audit logging designed in from the beginning. That early platform later matured into a gateway used for both internal and external APIs.
2021 – 2023 · Digital Equipment Capabilities
I then moved closer to the technology on John Deere machinery. On equipment licensing, I worked behind the customer experience for purchasing digital capabilities on equipment. A secure data-ingestion prototype I built during a hackathon showed that license data could be made useful to internal teams without giving up isolation or access controls, and helped make the case for a dedicated enterprise Data Engineering team.
2023 – 2025 · Enterprise Cloud Foundations
The next chapter moved from individual products to infrastructure used across the enterprise. I worked on cloud foundations serving engineering teams across John Deere. I led network consolidation, orchestrated a DNS migration, automated certificate renewal, and replaced manual account and VPC deletion with an event-driven workflow. The work reduced operational drag and made recurring infrastructure tasks easier to perform consistently.
The throughline across those chapters is the same instinct that now shapes my AI work: make the hard, important thing also the easy, default thing. Secure access that does not need to be reinvented. Compliance that does not depend on an engineer remembering it. Renewals that do not need a human in the loop. Onboarding through reproducible environments.
I’m not interested in pulling humans out of the loop. The target is complementarity: higher confidence, higher quality, lower cognitive load on the person doing the work.
That was true when the output was a pipeline nobody had to think about. It’s just as true now that the output is an agent.
Education & selected credentials
- B.S. Software Engineering, Iowa State University (Cum Laude, 2020). Cyber security minor.
- AWS Certified DevOps Engineer – Professional · May 2026
- AWS Certified Advanced Networking – Specialty · Aug 2025
Earlier certifications: SysOps Administrator – Associate (Feb 2025), Solutions Architect – Associate (Mar 2024), and Cloud Practitioner (Nov 2021).
Skills
Colophon
Built with Astro, Tailwind, and MDX; hosted on GitHub Pages. The source is public; see the link in the footer.