Self-Hosted Media Platform

Jan 2026 — Present UI / Frontend Systems Engineering

A private media archive and player with no CMS, no database, and no rented media stack.

I built this because I wanted a media system I could understand end to end. The filesystem decides what exists. JSON describes it. Caddy, PHP, and vanilla JavaScript do the rest. The result is a private media server that can browse, filter, and stream a mixed library of films, shows, and audio without a database, subscription, or heavyweight framework.

System shape

  1. Filesystem media library
  2. Ingest script
  3. Structured JSON catalogs
  4. PHP-rendered pages
  5. Client-side filtering and browsing
  6. Custom media player
  7. Caddy session auth and protected streaming

What I built

  • An ingest script that compares the files on disk to structured JSON catalogs, pulls OMDb metadata, and keeps JSON writes in one place
  • Two browsing modes: metadata cards for films and shows, plus a recursive tree browser for collections that do not fit that model
  • A custom player that handles audio and video through one interface, including HTTP range requests for video seeking
  • Session auth at the Caddy layer, covering pages, JSON indexes, poster images, and raw media files
  • Client-side filtering that loads sources once, keeps state in the DOM, and avoids re-fetching on every filter change

Infrastructure

  • Linux, Caddy, PHP-FPM, Cloudflare

What this demonstrates

  • JSON-backed UI rendering without a database dependency
  • Client-side filtering, browsing, and state management over structured content
  • Custom playback controls across audio and video
  • Protected asset delivery across pages, indexes, images, and raw media
  • Frontend implementation shaped by real infrastructure constraints