Research updates can break record links or detach images from their source.
Use stable IDs, persistent URL mappings, generated pages, and record/image-join checks.
← Selected work
Catalog & data systems
MJInserts
Turning structured research into a browsable catalog with stable identities, traceable records, and carefully framed probability tools.
MJ / INSERTS
Original catalog illustration · no third-party card imagery
THE PROBLEM
What needed to change
A collector catalog needs more than a gallery. Records must stay identifiable as research changes, permanent pages must remain stable, and probability information needs clear assumptions.
MY CONTRIBUTION
From concept to implementation
My approach connects the browsing experience to the data underneath it: preserve record identity, retain uncertainty, and make repeatable publishing possible without rebuilding the catalog by hand.
ARCHITECTURE
How the pieces fit
- Local structured research supplies records and source metadata.
- Python validates and joins records with image metadata using stable IDs.
- Node generators produce static catalog, collection, and detail pages.
- Lightweight browser JavaScript adds filtering, dialogs, and probability interactions.
ENGINEERING DECISIONS
Constraints and solutions
Missing or estimated odds can appear authoritative in a polished interface.
Preserve unknown values and explain the independent-pack and constant-rate assumptions behind modeled probabilities.
Large imagery can make a static site expensive to load.
Use responsive derivatives, reserved image dimensions, selective loading, and local font delivery while preserving original assets.
RESULTS
What the work delivers
- Launched a static catalog with permanent detail pages and collection navigation.
- Implemented repeatable publishing from structured records.
- Added searchable browsing, probability tools, and explicit image-provenance treatment.
Project records document checks for stable IDs, image joins, probability boundaries, generated pages, and keyboard dialog behavior. Historical performance measurements are scoped lab observations, not current field-performance guarantees.
ACCESSIBILITY
Where human evaluation matters
The implementation includes labeled filters, result announcements, native dialogs, focus restoration, reduced-motion support, and a static archive that can be browsed without JavaScript. Comprehensive conformance has not been established.