Why I Built an AI to Answer Recruiter Questions

The Recruiter Call That Started Everything

I got off a call with a recruiter and realized I had a problem. They asked me to send over my portfolio. I had one — but it was from when I first got into development, before I even landed my first job. That was years and multiple careers ago. It didn't represent anything I'd done since.

That's the portfolio problem every developer knows: you build it once, it goes stale, and by the time someone actually asks for it, it's a time capsule from a version of you that doesn't exist anymore.

I decided to fix it — not by updating a resume, but by building something that couldn't go stale.

What If a Recruiter Could Just Ask?

Here's the idea: instead of a static portfolio that goes out of date, what if there was an AI on the site that knew everything about my career? My work history, my tech stack, the projects I've built, my writing, my story. And a recruiter could just ask it questions.

On the right side of this page, you'll see an AI sidebar. It's open by default — it's not hidden behind a settings menu or a "chat with us" widget. It's front and center because that's the point.

Instead of hunting through pages or downloading a PDF, a recruiter can just ask:

"What's his experience with Claude and LLMs?"
"Has he worked at scale?"
"What did he build at CloneForce?"
"What's his story — how did he get into software?"

And get a real, conversational answer. Not a keyword match. Not a chatbot script. An AI that actually knows my career and can talk about it.

Built in a Day, Without Writing Much Code

Here's the part that matters most: this entire site — the blog engine, the design system, the AI sidebar, eight blog posts, the Experience page — was built in a single day.

I didn't write much code by hand. I designed the system, made architectural decisions, chose the tech stack, planned the structure, and guided Claude Code through chat. The actual code changes happened through conversation, not a text editor. I reviewed everything, made CSS adjustments, edited copy, and made sure it all worked together. But the implementation was AI-driven.

That's not a shortcut. That's the whole point.

Everything I write about on this blog — planning before coding, harness engineering, the Ralph Wiggum loop — this site is a live demo of all of it. I didn't just write about these patterns. I used them to build the thing you're reading right now.

The Stack

Framework: Next.js 14 with App Router
Blog engine: MDX with gray-matter, auto-generated TOC, reading time, related posts
Styling: Tailwind CSS with Inter font, minimal design system
AI sidebar: Powered by Open Notebook — an open-source AI knowledge base fed with my resume, projects, and writing
Deployment: Vercel
Images: Sharp for blur placeholders, custom optimization pipeline

The AI sidebar gets smarter every time I publish a new post. The knowledge base updates automatically. The portfolio never goes stale because the AI always has the latest context.

Why This Matters

This isn't just a portfolio feature. It's a proof of concept for how software gets built now.

A senior engineer with the right workflow and AI tools can design, build, and ship a production site in a day — not because AI writes code for you, but because the bottleneck was never typing. It was always decision-making, architecture, and knowing what to build. AI removes the mechanical work. The thinking is still yours.

If I can build an AI that knows everything about one person's career and answers questions conversationally — that's the same skill that builds AI-powered documentation systems, customer support tools, internal knowledge bases, and clinical decision support. The domain changes. The pattern doesn't.

The Real Portfolio

My real portfolio isn't this website. It's the fact that this website exists and I built it in a day.

It's not about the code. It's about the decisions — what to build, how to structure it, when to let AI handle implementation, and when to step in and guide. That's the job now. That's what I've been doing at every company I've worked at, from building concept cars at Honda to building AI pipelines at CloneForce.

The tools changed. The thinking didn't.

— Bill John Tran