01 · 5 min read
A script that finishes the scan before a human opens the first channel
Scanning 40+ channels a day, scoring every upload on six signals, and shipping a ranked report before the news cycle that made it relevant has closed.
Case studies
Every pipeline I ship has a public, genericized repo. These are the write-ups behind the hardest ones: the problem, the design decision that actually mattered, and the platform quirk that would have quietly broken it.
Anonymized on purpose: the channels are real, the client is not named, and every number here is one already stated in the linked repo's own README.
01 · 5 min read
Scanning 40+ channels a day, scoring every upload on six signals, and shipping a ranked report before the news cycle that made it relevant has closed.
02 · 4 min read
A YouTube channel tracker that refuses to hand over a single row until a second, separately-written pass of its own pipeline reaches the same answer from a fresh fetch.
03 · 4 min read
A batch rewrite of titles, descriptions, and tags across a whole YouTube playlist, authorized by the channel owner without handing over their login, and proven after the fact by reading the platform back, not by trusting the script's own log.
04 · 5 min read
A pipeline that turns a long bilingual livestream into short-clip candidates found that forcing a transcription model into one language doesn't make it fail on the other, it quietly translates over it, which is why every candidate clip gets re-checked before an editor ever sees it.
More landing soon
Leave your email. One short note when a new write-up lands here, nothing else.
Just your email, stored on my own private database. Never sold, never rented. Reply to any email to unsubscribe.