hn who's hiring — uk startups
i run this monthly to find early-stage uk startups actively hiring — companies i want to work with.
the problem
hn who's hiring turns hacker news' monthly hiring thread into a curated database of early-stage uk startups. the problem it solves: those threads are a goldmine of early-stage hiring intelligence, but filtering thousands of posts for uk roles by hand — then checking each company's funding stage — takes hours every month.
the approach
- 01 fetch
the pipeline pulls hacker news' current "who's hiring" thread in full.
- every comment in the monthly thread is retrieved through hacker news' api.
- 02 pre-filter
non-uk posts are removed before any ai runs.
- a keyword filter drops posts with no uk signal up front.
- this happens before the ai step, so the expensive part only ever sees relevant posts — a real cost cut, not a rounding error.
- 03 extract
claude reads each remaining uk post and structures it.
- it pulls company name, open roles, tech stack and location out of free-text hiring posts.
- 04 funding check
each company is checked against its funding stage — early-stage only.
- an apify crunchbase lookup returns each company's stage; series b and later are filtered out.
- companies with no crunchbase record are kept by default — for an early-stage startup, a missing record is usually a signal, not noise.
- the lookup only runs in the first half of the month, which roughly halves the apify cost without losing coverage.
- 05 log
results are written to a google sheet, newest first.
- rows are deduplicated by hacker news comment id, so re-running the pipeline never doubles an entry.
engineering challenges
- spending budget only where it pays
a hiring thread is thousands of posts, most irrelevant — and every ai call and crunchbase lookup costs money.
- run the ai over every post and most of the spend goes on posts that were never uk roles.
- look every company up on crunchbase every day and you pay for the same lookup over and over.
the pipeline filters before it spends, and spends on a schedule.
- a cheap keyword pre-filter removes non-uk posts before the ai step ever sees them.
- the crunchbase lookup runs only in the first half of each month, cutting that api cost roughly in half with no loss of coverage.
- early-stage when the data is missing
the feed is for early-stage uk startups — but "early-stage" isn't stamped on a hiring post.
- a post says who's hiring, not what funding stage the company is at.
- and many genuinely early companies have no crunchbase record at all.
funding stage is resolved externally, and a missing record is read correctly.
- an apify crunchbase lookup supplies the stage; series b and later are filtered out.
- a company with no crunchbase record is kept, not dropped — for early-stage startups, absent data is itself a signal.
the workflow
