# MipYip — Alex van Rossum (Full) > Fractional CTO and AI Systems Architect. I design the governance, infrastructure, and agent systems that make AI operationally trustworthy. This site documents technical case studies, the Pass@1 governance-first AI development methodology, open-source macOS tools, and a blog on AI governance, cognitive offloading, and systems thinking. ## About the Author Alex van Rossum is a Fractional CTO and AI Systems Architect based in the Atlanta metro area, Georgia. He specializes in technical leadership for organizations that have outgrown their existing engineering capacity — designing the governance, infrastructure (AWS, Kubernetes, EKS), and agent systems that make AI operationally trustworthy at scale. He also works in Salesforce architecture, vendor evaluation, crisis-response infrastructure recovery, and macOS application development. He developed Pass@1, a governance-first AI development methodology where architecture documents, coding standards, and constraint boundaries make AI agents productive session after session. - Site: https://mipyip.com - Blog: https://mipyip.com/blog - LinkedIn: https://linkedin.com/in/alexander-v - GitHub: https://github.com/avanrossum ## Key Pages - [About](https://mipyip.com/about): Background, expertise, and approach - [Approach](https://mipyip.com/approach): How Alex works — methodology, tools, philosophy - [Case Studies](https://mipyip.com/work): Technical deep dives into real projects - [Blog](https://mipyip.com/blog): Writing on AI governance, cognitive offloading, ADHD-aware systems - [Lab](https://mipyip.com/lab): Experimental projects and tools - [Contact](https://mipyip.com/contact): Get in touch ## Software Products - [SideMark](https://mipyip.com/products/sidemark): Free, open-source markdown editor for macOS built for AI co-editing workflows. Three-way merge, per-hunk diff, git gutter, auto-save. MIT licensed. - [Actions](https://mipyip.com/products/actions): macOS menu bar quick-action launcher. Scripts, shortcuts, links, folders — one click away. Coming soon. - [Panoptisana](https://mipyip.com/products/panoptisana): Open-source Asana visibility tool for macOS. Tasks, projects, inbox from the menu bar. ## Case Studies - [Sysadmin Claude](https://mipyip.com/work/sysadmin-claude): An autonomous infrastructure agent that SSHes into production servers, runs kubectl against a live EKS cluster, and gets faster every session through self-maintaining institutional knowledge. 9 servers, 150+ K8s deployments, 8 months in daily production use. - [Kubernetes Migration](https://mipyip.com/work/kubernetes-migration): The control plane died, SSH keys lost, certs expiring. Built a new EKS cluster and migrated 260 production deployments solo, zero downtime. Certs expiring in 3 weeks. Built a new EKS cluster from scratch. - [AWS Governance](https://mipyip.com/work/aws-governance): An AI agent audited years of AWS drift — cut $954/mo (30.9%) in 5 working days with zero production incidents. Read-only-first methodology, 100+ resources remediated, full documentation as byproduct. - [Salesforce Architecture](https://mipyip.com/work/salesforce-architecture): 32,000 metadata changes across five integrated systems. Delivered what three years of prior effort couldn't. Three years of stalled progress. Delivered by someone who had never used Salesforce before. - [AI Sprint Management](https://mipyip.com/work/ai-sprint-management): A human-in-the-loop AI agent that cut daily PM overhead from three hours to under one. Without sacrificing auditability or trust. - [Actions Development](https://mipyip.com/work/actions-development): 24 beta releases in 18 days, built solo with Claude Code under governance-first methodology. Architecture migrations, security audits, and a freemium business model — all built by one person. - [Agent Mail](https://mipyip.com/work/agent-mail): Async coordination between AI agents using Git as the mail server and Markdown as the message format. No server, no database, no API. - [Life Roadmap](https://mipyip.com/work/life-roadmap): Enterprise architecture applied to personal life management — version control, autonomous agents, separation of concerns. Built for an ADHD brain. ## Methodology **Pass@1** is a development methodology where governance documents are thorough enough that AI generates correct implementations on the first attempt. Not by being careful — by removing the ambiguity that causes failures in the first place. The governance documents — ROADMAP.md, ARCHITECTURE.md, CLAUDE.md, CHANGELOG.md — give AI agents persistent memory across sessions and encode institutional knowledge that survives context window resets. ## Blog Posts (Published) - [The Green-Light Problem](https://mipyip.com/blog/the-green-light-problem) (2026-03-22): The most expensive mistake in platform migrations isn't choosing the wrong platform. It's recommending one before you've validated the assumptions the recommendation depends on. A green light with unresolved checkpoints isn't a recommendation — it's a liability. - [SideMark: Building a Markdown Editor for Two Authors](https://mipyip.com/blog/sidemark-two-author-markdown) (2026-03-19): Every markdown editor assumes one author. When AI agents became co-editors, the 'file changed on disk' dialog became a workflow killer. SideMark is what happens when you solve that problem from the inside. - [The Costume Change Problem](https://mipyip.com/blog/the-costume-change-problem) (2026-03-16): Switching your AI between 'CEO mode' and 'engineer mode' produces better output than one generic assistant. But when every role shares the same context window, review becomes self-review in a different hat. The real question isn't what persona your AI should wear — it's what your AI shouldn't know when it evaluates this work. - [I Have a Team Now — It Just Happens to Be AI](https://mipyip.com/blog/i-have-a-team-now) (2026-03-14): The progression from 'AI helps me write emails' to 'I manage a department of specialized agents across six domains' happened in eight months. The output scaled. The cognitive load dropped. That combination is new. - [Cognitive Property: Who Owns the Way You Think?](https://mipyip.com/blog/cognitive-property) (2026-03-09): Your AI governance frameworks and decision-making logic are repeatable, transferable, and extractable. That's not a productivity feature. It's cognitive property. - [Cognitive Offloading](https://mipyip.com/blog/cognitive-offloading) (2026-03-05): Every system I tried — notebooks, apps, even a custom AI tool — solved the capture problem but failed at retrieval. Cognitive offloading is the methodology behind it all: deliberately choosing what stays in your mind and building systems to handle the rest. - [I Manage AI Agents the Way I Manage Teams](https://mipyip.com/blog/managing-agents-like-teams) (2026-03-04): The management principles that make human teams effective — separation of concerns, clear guidelines, professional focus, knowing when to restructure — apply directly to AI agent management. This isn't a metaphor. It's an operational framework. - [The Governance Documents](https://mipyip.com/blog/the-governance-documents) (2026-03-04): ROADMAP.md, ARCHITECTURE.md, CLAUDE.md, and CHANGELOG.md aren't project management overhead. They're the system that gives AI agents persistent memory across sessions — and the mechanism that makes Pass@1 work. - [Governance Is Architecture](https://mipyip.com/blog/governance-is-architecture) (2026-03-03): Most people treat AI governance as a compliance problem — policy documents, review committees, usage guidelines applied after the system is built. That's backwards. Governance is an architectural decision, and it needs to be designed in, not bolted on. - [I Built a Markdown Editor Because Every Other One Annoyed Me](https://mipyip.com/blog/simple-markdown-editor) (2026-03-02): Every markdown editor either wants to be Notion, wants your files in their cloud, or wants a subscription. So I built one that just edits markdown files on your computer. Open source, free, done in a weekend. - [AI Adoption: The 0.04% Don't Know They're the 0.04%](https://mipyip.com/blog/the-004-percent) (2026-03-01): The people actually building with AI are too busy building to post about it. If the LinkedIn AI feed makes you feel behind, you're measuring against the wrong cohort. - [The AI Perimeter: Where Automation Should End and Judgment Should Begin](https://mipyip.com/blog/when-not-to-use-ai) (2026-02-25): AI is the most powerful tool most of us have ever had. That makes knowing when not to use it the actual skill. - [LLMs Are Practically ADHD](https://mipyip.com/blog/llms-are-practically-adhd) (2026-02-24): ADHD and LLMs share the same failure modes — context loss, confabulation, drift without governance. The coping architectures for one transfer directly to the other. - [Your Reminders Don't Work Because They're Too Predictable](https://mipyip.com/blog/nag-bot) (2026-02-23): Traditional reminder apps fail ADHD brains because predictability breeds dismissal. I built a Telegram bot with fuzzy scheduling, natural language, and a relentless nag mode. - [Auto-Compaction Is Costing You Sessions](https://mipyip.com/blog/auto-compaction-is-costing-you-sessions) (2026-02-22): Claude Code auto-compacts at ~83% context usage. You lose the chance to control what's preserved. I built a script that warns you before it happens. - [Your AI Builds the Code. Who Reviews It?](https://mipyip.com/blog/the-adversary) (2026-02-22): If the same AI agent that writes your code also reviews it, you have someone grading their own homework. I built an adversarial code review agent to fix that. - [What Is Pass@1?](https://mipyip.com/blog/what-is-pass-at-1) (2026-02-21): Pass@1 is a development methodology where governance documents are thorough enough that AI generates correct implementations on the first attempt. Speed is a byproduct. Governance is the product.