Introducing AI Labs — Learning in Public

I've been heads-down for a while now. Between getting Leaderboard Fantasy to a point where it's steadily picking up users, putting TutorPro through real-world testing with my wife — who happens to be the toughest product tester I know — and navigating some personal family things, the past few months have been a lot. The kind of "a lot" that makes you want to disappear into a terminal for a few hours and just build something.

So that's exactly what I did.

Today I'm launching AI Labs — a personal portfolio of AI experiments, each one designed to explore a different corner of the AI engineering landscape.

Why Another Side Project?

Honest answer? I needed an outlet.

My two main platforms have reached a level of stability that I haven't had in a while. Leaderboard Fantasy is slowly growing its user base. TutorPro is in active user testing — and by "user testing" I mean my wife is giving me very direct feedback before we release it into the wild. Both products are in a good place, and for the first time in months, I have some breathing room.

My family would probably tell you I'm obsessed with this stuff. And there may be a little truth to that. But if I'm being honest, sitting down with a new AI pattern or an unfamiliar framework is a bit of an escape for me. It's the thing I reach for instead of doom scrolling. There are worse coping mechanisms.

More importantly, I realized I was getting comfortable. Comfortable with the tools I already knew, the patterns I'd already proven. And comfort is the enemy of growth — especially in a space that moves as fast as AI does right now.

What AI Labs Actually Is

AI Labs is a collection of standalone projects, each one targeting a different AI engineering discipline. Every project includes a live demo you can actually interact with and links to the GitHub repositories so you can dig into the code yourself.

Here's what's in the pipeline:

Classroom Clarity RAG

The first project out of the gate — a semantic Q&A API for school documents — was launched last night! Think handbooks, curriculum maps, policy documents. Upload them, ask questions in plain language, get grounded answers with source citations. Built with Spring Boot, Spring AI, PostgreSQL with pgvector, and Vertex AI. This one hits close to home given my wife's world in education.

Synthetic Student Generator

A tool that generates diverse student work samples based on rubrics — useful for teacher calibration exercises. This explores structured output generation and persona-consistent prompting with FastAPI and Gemini.

Diagram-as-Code Architect

Automatically generate Mermaid.js and SVG architecture diagrams from Spring Boot code and Terraform configs. A developer productivity play that uses LLMs for code-to-documentation translation.

Standard-Specific Fine-Tuner

The deep end of the pool — fine-tuning a small open model (Gemma 2B) on North Carolina education standards using PEFT/LoRA on Vertex AI. This is where I push past API consumption into actual model engineering.

The Point Isn't Perfection

I want to be upfront about something: these projects won't be perfect. Some of them will be rough around the edges. The UIs might be minimal. The error handling might be optimistic. And that's entirely the point.

This isn't a product launch. It's a learning journal with working code.

Each project is an excuse to go deep on something I haven't done before — whether that's vector embeddings and RAG pipelines, parameter-efficient fine-tuning, or building real-time classification systems. I'm deliberately stretching across languages (Java, Python, JavaScript), cloud services (Cloud Run, Cloud Functions, Cloud SQL), and AI patterns (RAG, structured generation, fine-tuning, multimodal classification) because breadth matters just as much as depth when you're trying to stay sharp.

An Invitation to Tinker

The reason I'm putting all of this out in the open is simple: I want other people to learn alongside me. I really enjoy that some of you have already reached out to learn more.

Every project will have a live site where you can test things out and a GitHub repo where you can fork, clone, and make it your own. Use them as starting points. Break them apart. Rebuild them differently. Or just read through the code to see how someone else approached the problem.

If you've been following my writing, you know I believe the best way to learn AI engineering is to build things — not watch tutorials, not read whitepapers, but actually get your hands dirty with real projects that solve real problems. AI Labs is me practicing what I preach.

Here's the thing I keep coming back to: curiosity isn't a phase. It's in our DNA. That pull you feel when you see a new technology and think "I wonder how that works" — that's not a distraction, that's the signal. The moment you stop following it, you stop growing. It doesn't matter if you've been in this industry for 3 years or 30. The people who stay sharp are the ones who never stop being students.

This is what the web was supposed to be about — experimentation, sharing knowledge, building on each other's work. Before the algorithms and the paywalls and the engagement metrics, there was just people putting stuff out there because someone else might find it useful. That's the energy I'm going for here.

What's Next

Classroom Clarity RAG is the first project going live, and I'll be writing deeper technical posts as each experiment matures. If you're interested in any particular project or AI pattern, let me know — I'm genuinely building this in the open and happy to adjust based on what people want to learn about.

You can check out the full portfolio at labs.jking.ai, and all the source code lives on GitHub.

Now if you'll excuse me, I have lots to learn.

–Jeremy