HN Match

HN Match is a free matching site for the two monthly hiring threads on Hacker News. Every month the community posts a Who is Hiring thread full of job comments and a Who Wants to Be Hired thread full of candidate comments, and the two never meet except by hand. HN Match reads both, extracts structured data from each comment, and scores every plausible pairing so a job can see its likely candidates and a candidate can see the jobs worth a reply, with the reasons for each score laid out. It's a small, focused tool, and the site doesn't pretend otherwise.

The problem is scale and format. Each thread runs to hundreds of free-form comments in whatever structure the poster chose, some with a tidy list of skills and a salary range, others with three paragraphs of prose and no location. Reading them all is a day's work, and searching for keywords misses the posts that describe the same role in different words. Companies get buried and candidates never see the one job that fits, purely because both were written in prose by people in a hurry. The mismatch runs both ways. A candidate who writes remote only, senior, and Rust in one line and a company that writes the same requirements across three paragraphs will never find each other with a text search, even though the fit is obvious once both are structured.

The pipeline is described openly on the site's Method page. It pulls both threads through the public Hacker News API and runs automatically on the third of each month, which usually means it reflects the threads posted on the first. A language model extracts skills, location, remote preference, salary range, seniority, industry domains, and visa requirements from each comment. Hard filters then remove impossible pairs, such as remote mismatches, salary ranges that don't overlap, and visa needs a company can't support. Each remaining pair gets a weighted score from 0 to 100, with skill overlap worth 40 percent, domain fit 20, seniority alignment 20, salary overlap 10, and a base floor of 10. Missing fields aren't penalised, they're handled gracefully, so a terse post isn't buried just because it left out a salary. The hard filters are the part that keeps the scores honest. A perfect skills match is worthless if the job is on-site in one country and the candidate needs a visa the company won't sponsor, so those pairs never reach the scoring stage at all.

The result is a browsable site with jobs, candidates, and pre-scored matches broken down by signal, so you can see why a pairing scored the way it did rather than trusting a number. The September 2026 edition listed 105 jobs and 130 candidates, each with salary ranges, locations, and required skills where the original post included them. There's no sign-up and nothing to submit, since the data comes straight from the threads, and the navigation is just browse, jobs, candidates, and the method page. Each job page carries the salary range, location, and technologies the poster mentioned, and the match list shows the signal breakdown, so a 72 with strong skill overlap but weak seniority reads differently from a 72 built the other way.

It's for engineers who post in Who Wants to Be Hired and want a ranked shortlist instead of scrolling, and for hiring managers who post in Who is Hiring and want to know which candidates in that month's thread actually overlap with the role on skills, seniority, and pay. It's also useful for anyone curious about the shape of the HN job market in a given month, since the extracted fields make the threads comparable in a way the raw comments never are. Recruiters who mine the thread each month will find it does the first pass they'd otherwise do by hand.

What makes it different from a job board is that it doesn't host jobs at all. It's a lens on threads that already exist, with the scoring method published and the caveats stated. The site is clear that extraction is automated, that terse posts may parse imperfectly, and that scores measure plausibility of fit as conversation starters, not hiring decisions. The limits follow from the design. It's a monthly snapshot rather than a live feed, it covers only the HN threads, the site doesn't name its builder or offer a contact, and there's no way to correct a bad parse of your own post other than writing it more clearly next month. The weights are also fixed. You can't tell it that salary matters more to you than domain, so the ranking reflects the site's priorities rather than yours.

Access is free with no account, no pricing, and no sign-up. The site doesn't publish a source repository or a contact email, so it's best treated as a community utility that refreshes once a month, and the way to use it is to find your own post or role in the current edition, read the matches, and treat the score as a reason to open a conversation rather than a verdict. Because everything comes from public comments, there's nothing to delete or opt out of here that isn't already on Hacker News.