Skip to content

Beta.applyNchill is in active testing — expect rough edges, and check what it sends before you rely on it.

six stages · apply-service · port 8083

Match

Ranks the enriched corpus against your profile. Gates remove; they never quietly demote.

How this stage flows

Structured jobs from the shared corpus and your own profile go through two gates that remove rather than demote: an experience gate and a role gate. What survives is ranked on four keys in fixed priority — experience fit, source tier, skill overlap, recency — into a queue that belongs to you alone.

gates first — they remove rather than demote

then ranked, four keys in fixed priority

per user, from one shared corpus

One corpus, many rankings. The jobs are shared; every list built from them belongs to exactly one person.

Ranking is four keys in a fixed order rather than one blended score. A blended score is easy to write and impossible to argue with: when a bad job comes third, you cannot tell which term did it. In a fixed priority you can.

Gates remove, they do not demote

A role that wants twelve years when you have four is not a weak match, it is not a match. Scoring it down leaves it in the list to be reached on a quiet day. Removing it is the honest operation, and it is the difference between a queue you trust and one you re-check by hand.

Shared corpus, private lists

The jobs are swept once and shared, because a posting is a public fact and sweeping it twice is rudeness to the board. Everything derived from it is per user: the ranking, the queue, the deck, the ledger. Those are separated by user in the query itself, not filtered afterwards.

That separation has been the source of every cross-user leak worth having. A list built without the user in the query looks completely correct until there are two accounts.

Found this useful? Add applyNchill as a preferred source on Google