Reel Automation
An end-to-end Golang media automation pipeline that monitors subreddits via OAuth2, scores viral video velocity, transcodes 9:16 vertical video with FFmpeg, and publishes to Instagram Graph API.
Autonomous Viral Content Engineering
Reel Automation is a modular Go system (cmd/reel-automation) designed to detect, transform, and publish trending short-form video content with zero human intervention.
Triggered every two hours by robfig/cron, the Reddit OAuth2 fetcher pulls candidate video posts across configured subreddits and filters out NSFW, locked, or stale threads.
Velocity Ranking, Deduplication & FFmpeg Transcoding
Instead of relying on raw upvote counts, the ranking engine computes upvote velocity (upvotes / age_in_minutes) and ranks candidates via score = velocity * 0.7 + comments * 0.3.
Unseen posts pass through sanitized yt-dlp downloading and an FFmpeg 9:16 vertical formatting pipeline before publishing to the Instagram Graph API with configurable exponential backoff retries.
Fault-Tolerant Worker Design
Every stage—from download to transcode and Graph API container publish—is guarded by stage-specific retry budgets and persistent PostgreSQL deduplication.




