Pet content pipeline
An Instagram account for a cat, run by a pipeline that generates, brands and queues the work — a person approves every post, and the ledger says whether any of it landed.
-
Nobody is awake when it runs.
A timer starts it. It generates the image and the video for the day and assembles the caption from a content engine. Nobody approves anything yet — that gate comes later, before a post exists.
-
The template is a system, not a habit.
Brand frames applied the same way every day, and a workflow doctor that fails the run when any script or scheduled step goes missing from the documented diagram.
-
Then it checks whether any of it worked.
Post counts, median views and engagement per platform, appended to a ledger that only ever grows, and served on a dashboard rather than guessed at from the app.
- 01 Built an end-to-end automated content pipeline — asset generation, branded templating from a design system, caption and hashtag authoring, and a human approval step, delivering the finished asset to a phone album to be posted
- 02 Designed a weekly planning run in which several agents propose the coming week's content and each proposal arrives as an approve/edit/reject card, so a person approves every item and a worker builds only what was approved
- 03 Instrumented the pipeline with an append-only performance ledger behind every post and comment, so what gets made next is decided from measured results rather than taste
- 04 Runs unattended on scheduled services, guarded so that a checkout which is not the live one cannot consume the approval taps — they are acknowledged server-side, so a stray run would swallow a press the real tree then never sees
Interface recreated with invented data — no production screenshot is published. Private system; walkthrough and code review available on request.
- Role
- Sole engineer
- Timeline
- 2026 — ongoing
- Stack
- Python · generative media APIs · cron · systemd