Evolution · The connections between projects

One project leads
to the next.

Each project opens up another question. These are the threads running through the portfolio: the ideas that carried forward, the tools that changed and the findings that shaped each new version.

Four threads through the work

Healthcare

2025–2026

From finding the data to helping someone use it.

The healthcare work began with a practical gap: useful information was spread across directories, websites and spreadsheets. Building a consistent register led to demand and cost models. Those models then needed better ways to explore a place, compare a scenario and explain the result.

  1. 01

    Build the foundation

    Service discovery, clinic records and geographic joins.

  2. 02

    Ask better questions

    Workforce estimates, procedure costs and scenario workbooks.

  3. 03

    Make it explorable

    BusiIndex, site-screening maps and repeatable research reports.

The models and the interfaces grew together. A clearer interface often revealed a data definition that needed another look.

Creative tools

2026

Keep working after the first draft.

InterfaceGenie, later called Lunar, explored a web-based design tool. The report and presentation work raised a question that carried into Aphelion: how do you keep an AI-generated page as editable and recoverable as one made by hand?

  1. 01

    InterfaceGenie / Lunar

    Explore document creation through a web editor.

  2. 02

    Carry the question forward

    Focus on editable pages, layout and the history of a change.

  3. 03

    Aphelion

    A separate native editor with local generation and structured documents.

Aphelion is a new implementation carrying forward the product direction. Its editor and generation quality are still developing.

Wearable research

2026

Better questions start with better recordings.

CHARM is the wristband; AutoChore provided early collection and inspection tools. Ichne grew into the modelling system around them. The immediate research is about what the sensors recorded, how labels line up and what an activity classifier can learn from a small study.

  1. 01

    CHARM

    The wearable and its sensor recordings.

  2. 02

    AutoChore

    Collect, inspect and label the data.

  3. 03

    Ichne

    Quality checks, activity prediction and grouped evaluation.

Route reconstruction, floor maps and narrative reporting remain later research stages. The current work is building a stronger foundation for them.

Market research

2026

From a strategy idea to a decision you can inspect.

Early stock-analysis tools developed into several terminal and strategy experiments. Electron, Rust and later Windows desktop builds explored different parts of the problem. Chronos brings that history together as a research platform for examining signals, forecasts and paper-trading decisions.

  1. 01

    Analysis tools

    Explore market data and strategy ideas.

  2. 02

    Terminal experiments

    Develop charting, strategy engines and local model services.

  3. 03

    Chronos

    Inspect the decision path in a richer desktop research environment.

The related builds have different roles. Backtests and paper runs remain experiments, with their assumptions and evaluation limits attached.

Study → practice

Learning belongs
inside the build.

Actuarial and finance studies provide a foundation for questioning assumptions. The course books work through the mathematics, while practical builds expose the details that need further investigation. Research notes connect the two.

Where the work goes next

A useful question is a good starting point.

Explore the current projects, or start a conversation about something you’re trying to understand or build.