Mckay Wrigley Portfolio
Founder of Takeoff AI
Founder of Takeoff AI, which builds the MagicLearn AI courses and the not-yet-launched AgentShare workspace. The personal site is deliberately minimal: a one-line bio, links out to GitHub (Chatbot UI, 33.3k stars), and just two long posts, one on Claude Opus 4.5 (December 2025) and a July 2025 argument, backed by an Obsidian note-taking workflow, for renaming Claude Code as "Claude Agent."
Location
Didn't say. We're respecting the mystery.
Companies
Skills
Notable
- Chatbot UI, open-source AI chat app: 33.3k GitHub stars, 9.4k forks
- mckays-app-template, his full-stack starter template: 1.9k GitHub stars
- MagicLearn (Takeoff AI) reports a 4.9/5 rating from 25,000+ learners (per the site)
What makes it work
A breakdown of the choices that make this portfolio stand out.
Posts anchored to specific tools and versions
Writing about a named model version and a named tool, rather than "AI agents" in the abstract, shows the analysis comes from actually using the current generation of tools, not general commentary.
A specific, arguable technical opinion
The "Claude Agent" post argues Claude Code is really a general agentic system disguised as a coding tool, then backs it with a concrete non-coding setup: pointing it at an Obsidian vault for note-taking. A position plus a reproducible workflow is what makes a post worth discussing rather than skimming.
Founder identity paired with hands-on technical writing
Being both a startup founder and a hands-on technical writer signals the writing comes from someone building with these tools daily, not observing from the sidelines.
What AI Engineers can take from this
Specific, actionable tips to apply to your own portfolio — no generic advice.
Anchor your technical posts to specific named tools and versions, not abstract trends — it signals hands-on use over general commentary.
State an actual, arguable technical opinion rather than a neutral summary of a release — it's what makes a post worth discussing.
If you're building a product with the tools you write about, say so — direct, current usage is a credibility signal, not a conflict to hide.
Projects
Two posts, a one-line bio and a GitHub link: Mckay Wrigley's site leaves the showing to the code. The restraint works because Chatbot UI's 33.3k stars make the case on their own.
"The open-source AI chat app for everyone": a self-hostable ChatGPT-style interface for any model, with an official hosted version. Version 2.0 moved storage to Supabase (Postgres) with database migrations, and the 1.0 code is kept on a legacy branch.
It is a full-stack AI product at real open-source scale: auth and persistence, multi-provider adapters, a migration path for self-hosters, and issue triage for a large user base.
Model-agnostic chat UI with provider adapters (including Gemini)
Supabase-backed 2.0, with npm run db-push migrations for hosted instances
Keeps 1.0 on a legacy branch instead of breaking existing users
MagicLearn
Takeoff AI's course platform for learning to build with AI tools, rebranded from "Takeoff". Courses cover Cursor, Claude Code, AI agents, building apps with AI, prompt engineering and learning to code, aimed at people with no experience.
It shows the founder side of AI engineering: turning hands-on tool expertise into a product with a pricing page, a changelog and a real learner base.
Course catalog built around current tools (Cursor, Claude Code) rather than generic AI theory
Rebrand from Takeoff shown openly on the site
Aimed at learners with no prior experience
More AI Engineer portfolios
Ready to build your portfolio?
Follow the AI Engineer roadmap — skills, projects, and timeline to get hired.