LucidCairn
LucidCairn

BUILD · BREAK · LEARN · REPEAT

WILLIAM · DEVELOPER / TINKERER / CURIOUS HUMAN

Hi, I'm William.
I make things.

Some useful. Some just for fun.

Sometimes it's a problem from work. Sometimes it's just a tool I wish existed. I like seeing how far an idea can go.

See what I've made A little about me
KEEP EXPLORING / SCROLL DOWN

Build. Break. Learn. Repeat.

Weathered stones stacked as a quiet landmark.
01 / WAYMARKA stone, a path, a beginning.

02 / THE TERRITORIES

Things I've made.
And why I made them.

Some started with a problem at work. Others began with “I wish this existed.” Here's what happened next.

01 / RESEARCH & WRITING

Folio

Long reports. No starting over.

When a chapter fails, the whole report shouldn't have to start again. I built Folio around that idea.

Behind the build
LISTEN / NOTICE / SPEAK

02 / LANGUAGE LEARNING

LINGA

Go from watching to speaking.

Learn with videos you love. Catch a phrase, replay the moment and make it yours.

03 / MARKET RESEARCH

Quant Compass

Less noise. A clearer view.

A place to look at trends, risks and the market questions I keep coming back to.

04 / DATA EXPLORATION

Queryline

Ask questions. Find the right data.

The real challenge with hundreds of tables is knowing where to look. Queryline starts with a question.

Behind the build

05 / INTERVIEW PRACTICE

MockForge

Practice the thinking, not the script.

A place to rehearse technical questions, learn from feedback and take another shot.

06 / JUST FOR FUN

Seasonal Echo

Same temperature. A different day.

Spin through a year and find days with similar temperatures in another season or city. A little experiment with climate averages.

03 / UNDER THE SURFACE

The interesting part?
Making it hold together.

A clever answer is the easy part. I care about what happens when an agent needs to choose a tool, recover from a failed step, or pick up where it left off.

WHAT I'M EXPLORING

  • 01 / SYSTEMS Agent Harness Tool boundaries, execution state, and recovery.
  • 02 / CONTEXT Context Engineering The right context at the right moment.
  • 03 / KNOWLEDGE Agentic RAG Grounded retrieval and resumable long tasks.
  • 04 / RELIABILITY Evals & Tracing Make failures visible. Make progress measurable.
INSIDE A RUN / 01 One run. Four checkpoints.
RUN OBJECTIVE

Turn a complex request into a dependable result.

01
Scope Define outcomes, constraints and tool access
02
Plan Bound the next step and set checkpoints
03
Execute Retrieve, call tools and persist state
04
Verify Evaluate, recover and retry when needed
EXECUTION TRACE 01 / 04

A good harness makes each step inspectable — and failure recoverable.

04 / THE PRACTICE

I care about morethan getting it built.

The details that matter rarely announce themselves: how things connect, how they recover when something fails, and whether they still feel right on the tenth visit. Those are the things I keep refining.

A FEW RULES I KEEP

  • Make the complicated feel clear.
  • Test an idea before declaring victory.
  • Care about the smallest details, too.

THREE THREADS

  1. Agents

    Give complex tasks a next step — and a way back.

  2. Data

    Bring a good question closer to a useful answer.

  3. Products

    Grow small ideas into things worth keeping.

05 / THE HORIZON

What's next?
I'm curious, too.

If you're working on something interesting—or just like figuring things out—you'll find me on GitHub.

Find me on GitHub