Skip to content
The pipeline

People & referrers

Finding the humans who could refer you — without your accounts.

The product is a named person who works there now and could refer you. Finding them — without ever logging into your own accounts — is the heart of ZoNuLy, which is why referral work gets most of every cycle.

The goal per company#

Three ready referrers, at least one with a usable address. “Ready” means verified-current or likely-current, a person rather than an inbox, not on the do-not-contact list, and with some identity (name, GitHub or LinkedIn).

Where people are found#

  1. Founders and team pages
    YC's page lists active founders; the company's own team and About pages name people next to their roles — the strongest evidence of all.
  2. One Exa people search
    A single metered search per company for current profiles, inside a daily reserve.
  3. Public LinkedIn, keyless
    Search-result titles like “Name – Engineer at Company”, scoped to the company's city. Profiles are never fetched while logged in.
  4. GitHub
    Members of the company's GitHub organisation and recent committers to its repositories — engineers by definition, often with a commit address.
  5. Public mentions
    Talks, podcasts and blog bylines that name someone at the company.
  6. Apify People tab
    For high-paying companies still short of referrers: the company's LinkedIn People tab bought from Apify by its own job every 30 minutes — up to three companies a run, 25 people each, in the company's city — inside the daily cap of $0.60 (the $19 Starter credit spread over a month).

The rules#

City lock#

For a Bengaluru company only people based in Bengaluruare kept; for Hyderabad, only Hyderabad. Location means the location on their own profile — never a college, never a search snippet. A person with no known location is dropped under the lock. Apify receives the city as a server-side filter, so you don't pay for people in other countries.

Employment must be current#

A public profile that lists the company with no end date is verified-current; with an end date, the person is former and dropped. A commit alone is not employment. Referrers are re-confirmed after 90 days and downgraded if the evidence has gone.

Ranked by who can actually refer#

RankRole class
1Senior, staff, lead or founding engineer
2Software engineer, SDE 1–3
3Founder, CTO, co-founder
4Engineering manager
5Technical recruiter / tech HR
6General HR / talent
9Anyone else — never asked first

Since 23 Oct 2026 the tech team ranks first: a referral from an engineer on the team you would join carries more than one from a founder's inbox. The order lives in targeting.people.roles in config.yaml.

The actors layer#

jobhunter/actors/ is “our own Apify”: small programs (actors) with a run ledger, datasets, a page cache, and per-host politeness. LinkedIn is read only as a logged-out guest, at most one request every 20–45 seconds, and a circuit breaker stops every actor on a host the moment it sees a login wall, status 999 or 429.

find a company's people
python scripts/run.py actor people 2802 --linkedin https://www.linkedin.com/company/skylotech \
  --source apify --apify-max 25 --title engineer

Results are saved as contacts through the guardrails by default and appear on the company page and on Leads. Use --no-save to only look.

Your accounts are never used
No scraper logs in to LinkedIn, X or GitHub as you. A banned account would destroy the very network referrals depend on. Where a logged-in view is needed, Apify's own accounts carry the risk.