Looking for a job today feels like a second job. I'm an ML engineer, and I know the routine: search, open a posting, read it, work out whether you even fit, repeat. Oink is what I built to do that first pass for you. It sends a short, explained list of roles instead of an endless feed. It does not apply for you. It gives you the original link, and the decision stays yours.
1. You describe the job you want
A search brief works better than a bag of keywords. Oink can use your target role, skills, seniority, preferred work arrangement, location or work eligibility, salary floor, and explicit dealbreakers. Telegram onboarding starts from those answers and can add a CV.
Your wishes count as wishes. A past job on a resume shows what you can do, and it says nothing about whether you want to do it again. If your brief says "staff backend roles, remote in Europe, no ad-tech", a title from your work history should not quietly replace those constraints. When your stated target and your resume disagree, the target wins. We learned this the hard way: a typed search for senior product manager roles once returned an infrastructure job, because a stored engineering resume had overridden the request.
To write a brief, start with how to describe the job you want to an AI. It doesn't need to be nice prose. It needs enough detail to tell a good role from one that only looks plausible.
2. Oink searches more than one kind of source
Oink combines broad job boards, specialist and regional boards, employer career systems, monthly hiring threads, and open-web discovery. Some searches are shared across users and filtered later. Others, including LinkedIn-derived and open-web searches, use phrases built from your current profile.
Sources change. A provider can be slow, blocked, empty for your query, or switched off by us. So Oink never promises that every named source takes part in every run. The job-sources guide lists the categories and explains what can go missing.
Results are put into one comparable shape and deduplicated before matching. The original posting stays the authority: confirm that the role is still open and that its terms haven't changed before you apply. The guide to spotting stale and ghost listings shows the manual check.
3. Explicit conflicts go first, then the score
Some evidence is mechanical. A role that requires a location you can't work from, a work arrangement you ruled out, or a clearly different seniority shouldn't take the same attention as a close match with one unknown detail. Oink screens out those explicit conflicts, then scores the viable candidates with more context.
The score runs from 0 to 5, and a higher number means a closer fit to your brief. It ranks relevance. It is not a chance of getting an interview, and it can't tell whether the employer will consider you eligible. A detail the posting leaves out stays missing, and the card says so instead of inventing it.
The AI job-matching guide explains the scale and the reasons behind each score. If you have been applying at volume without replies, score before you apply is the manual version of the same idea.
4. The strongest matches become cards
On a schedule, Oink ranks the eligible results and sends a limited set of the best ones, above your score floor. A Telegram card can show the title, company, location, the original link, the score, concrete reasons for the fit, concerns or unknowns, and whatever structured details the posting had.
From the card you can open the posting, mark it applied, ask for a deeper fit analysis, or answer "more like this" or "not for me". Marking a role applied updates Oink's tracker and does not submit anything to the employer. Feedback counts as new evidence about what you want, and later scoring can use it. One tap does not retrain anything on the spot.
The same pipeline feeds the Oink iOS app and a linked Telegram account. Telegram is the quickest way to get the list without keeping a browser tab open, and the Telegram job-alert guide covers delivery and its limits.
What Oink doesn't do
- It doesn't promise that every source is healthy or returns results on every run.
- It doesn't promise that a third-party listing is current, legitimate, or still accepting applications.
- It doesn't turn a score into an interview probability.
- It doesn't auto-apply or make the decision for you.
- It doesn't replace checking work authorization, pay, and application terms at the original source.
My biggest mistake while building this was adding extra features, like resume advice, while scoring quality was still not great. Scoring is the main feature, and it is still being improved. Next time you get a card, read the reasons before the number, then open the original posting.
Describe the role you want and add a resume where required or useful. Oink searches the public sources available to your run, ranks viable postings 0–5, and delivers explained matches on the web or in Telegram.