Solo — Strategy, Design, Engineering, Deployment
This case study is not about a website. It's about what happens when a design leader removes the walls between disciplines and runs the whole product lifecycle himself. The website is evidence. The process is the point.
The Brief
Most designers outsource everything outside of design. This project was about closing those gaps.
The real brief
The surface problem: I needed a portfolio. The deeper problem: a static Squarespace site was actively contradicting the claim that I think in systems and ship things.
Every update required cloning a repo and manually editing code. For a site that needed to evolve as fast as a job search, that friction was a liability.
So this became two briefs. Build a portfolio that proves systems thinking. Then — post-launch — build the tools to keep it current without engineering overhead.
Phase 1
Before a single pixel, a content audit. Most designers skip this step. It's a discipline problem, not a tool problem.
Phase 2
Tokens first, components second. One-sentence brief: near-black, amber-accented, editorial — typographically bold where it counts, sparse everywhere else.
Phase 3 — v1.0.0
First launch shipped May 20. The design didn't stop at Figma — it went all the way to production.
This is where AI's role becomes explicit. Accelerant, not author. Every architectural decision was made by a human who understood the tradeoffs.
Post-Launch
Adding a project meant editing JSON files, running a local build, and pushing a commit. For a site evolving as fast as a job search, that's unacceptable.
Sanity and Contentful solve the right problem — but add dependency, cost, and abstraction. This was solvable in-house. Building it was itself a proof point.
JWT auth, protected routes, project editor with visual section builder, site content editor. One surface, no external services, no ongoing cost.
The Build Arc
Each version solved a specific problem. The direction never changed — only what was possible.
Content strategy, design system, full-stack build. Live on Vercel in ~2 weeks.
JWT auth, project editor, site content editor. No code required to update the portfolio.
13 section components. Visual card-based editor. Case study page: 800 lines → 110.
GitHub API writes. Every save commits to the repo. Vercel rebuilds automatically.
The CMS
The /admin interface handles the full content lifecycle without touching code.
Projects, sections, and site content are all editable from a single password-protected dashboard. Every save commits directly to the repo — no terminal, no deploy command, no engineering overhead.
What shipped
Phase 6 — v1.3.0
Vercel’s serverless runtime has a read-only filesystem. The right fix depended on which tradeoffs were worth accepting.
Reflection
Running the whole lifecycle myself is what lets me teach it. You can't turn a handoff into a repeatable pattern for your team until you've stood on both ends of it.
You understand the real cost of a late design change. You know which engineering decisions are genuinely constrained and which are just habit. You write better briefs because you've been on the other end of bad ones. This project was built with Claude as a collaborator throughout — content strategy, design direction, every line of code. The judgment calls — what to cut, what to build, what tradeoff to make — were human. AI accelerated every phase by removing the parts that don't require judgment. That's the integration model, not the replacement model. The walls between disciplines are expensive. This project is proof that they're also optional.