Methodology and data sources

Source

Every datapoint comes from Hacker News' monthly “Ask HN: Who is hiring?” threads, where companies post open roles as public comments. We read the top-level job posts, not replies. This is a beloved, public, primary source: the pay figures are what companies themselves wrote next to a role.

Extraction

Each job post is passed through a language model that extracts structured fields: company, role, salary range, currency, remote flag, seniority level, location, and tech stack. Posts without a stated salary are skipped. We keep the raw comment text so every figure is traceable to its original post on Hacker News.

Normalization

Salaries are read in the currency as written (USD, EUR, GBP, and others) and converted to USD using a quarterly reference rate for cross-comparison and medians. The native currency is always shown on the listing. Hourly rates are annualized. We normalize tech tags to canonical names and drop generic phrases.

Deduplication

Companies re-post the same role across months. We collapse duplicates with a content fingerprint (normalized company + role + salary) and keep the freshest version, so a repeat posting is counted once.

Limitations (read this)

• These are advertised ranges from job posts, not verified paid compensation. They skew toward the lower bound of real pay and toward companies transparent enough to post numbers.
• Hacker News skews senior, US-heavy, and toward certain stacks. Treat it as "startups that hire on HN", not "all startups".
• Small cuts (a single tech in one region) have small sample sizes; we show the count and avoid headline claims below ~30 listings.

License & citation

Aggregated figures are free to cite under CC BY 4.0. Please attribute: “Source: Hacker News Who is Hiring, aggregated by StartupJobList” with a link.