Jamieson Rothwell

@jeheecheon Thanks, Jehee! Really appreciate you picking up on those two design choices—keeping data strictly local and calling out ghost jobs were huge priorities from day one.

For ghost job detection, it’s a combination of a few key signals:

Posting Cadence & Persistence: Tracking how often a position gets silently pulled and re-listed without meaningful updates, or stays active far past standard hiring windows.

Description & Meta Patterns: Flagging structural template reuse, overly generic language designed just to farm resumes, or discrepancies between direct ATS boards and third-party aggregators.

Recency vs. Activity: Cross-referencing listing age against known hiring patterns to flag stale posts.

Combining that with local LLM evaluation means all that analysis happens right on your device without routing your pipeline data through a third-party server.

Out of curiosity, are there specific red flags you’ve run into in your own search that you wish trackers caught earlier?

is it possible to use this with native apps?It would be really valuable to run tests like this without having to build a bespoke solution

Ive used a tool similar to this and it was false listing jobs as eligible in my country that werent. How do you overcome that?

Interesting idea. Now that some of the larger players are rolling out internal tools that do some similar tasks how will you discover new areas for growth