Skip to content

All projects

2024 to 2026 · Archive · Investor relations

Everdime

A marketing and investor-relations site for a discontinued Web 3.0 and AI venture platform, its lost content rebuilt from a database dump and Wayback snapshots.

  • Next.js 14
  • styled-components
  • NestJS 9
  • TypeORM
  • MariaDB 10.7
database rows rebuilt
99
table the whole site reads from
1
metadata rows holding SEO as JSON
4
timeout on the string-storage call
5 s

What I walked into

A discontinued venture whose 2024 content was lost. What remained was a production database dump and June 2024 Wayback snapshots.

The entire site reads from one string_storage table whose keys encode layout position.


What I changed, and why

  1. Keys that say where they go

    Keys encode layout position, as in home_page_offer_section_1_title, and four metadata_<page> rows hold JSON documents, which made SEO itself editable content.

  2. Byte for byte

    Every recovered string was verified byte for byte in both the SQL seed and the TypeScript fallback.

  3. The original architecture, kept

    SSR still calls GET /string-storage?page= with a 5-second timeout, and every non-empty row the API returns overwrites the archived fallback string by string.

Every recovered string was verified byte for byte.

Where it landed

The site is back with its 2024 content rebuilt as 99 database rows.

It keeps the original architecture: the API wins whenever it answers, and the archived fallback fills in string by string when it does not.