jlm·technologies

06 · live

Hiring Platform

Apply for a job from any phone. No account, no CV, four languages.

4

languages, enforced at build time

3

apply paths, AI-optional

5

days to first live pilot

0

personal details visible to the ranker

/ interactive · toggle what the ranking model is allowed to see

try it, everything responds

candidate → ranking model payload

namenever sent
gendernever sent
agenever sent
racenever sent
photonever sent
phonenever sent
experience4 yrs: waitress, barista (Gardens, CT)
skillsPOS systems · stock counts · barista training
languagesisiXhosa, English, Afrikaans
availabilityweekends + evening shifts
references2 (count only, names withheld)

The whitelist is enforced in code, not in a prompt: name, gender, age, race, photo and contact details are stripped before the payload is built, reference names collapse to a count, and any ranking claim without cited evidence from the remaining fields is dropped. The score comes back with the evidence attached: a shortlist a manager can defend.

A note on this demo. Every figure, name and data point here is invented for illustration. No client data, real economics or production configuration appears anywhere on this site. What is real is the machinery: the calculations, the rules and the logic running in your browser are the same ones the production system uses, so the demo behaves exactly as the real thing behaves.

What it is

/ in plain terms

Someone looking for work as a waiter, driver or cashier can apply from any phone, in their own language, with no CV and no account. Employers get a ranked shortlist with the evidence for each candidate spelled out. The ranking system never gets to see anyone's name, gender, age, race or photo, so it cannot discriminate on them even by accident.

AI hiring for South Africa's operational workforce: waiters, drivers, cashiers, security staff. Employers post a role and receive a shortlist where every ranking claim cites its evidence. Workers apply by photographing a CV, speaking their work story aloud, or tapping through guided questions, in English, isiZulu, isiXhosa or Afrikaans, verified by SMS. Built in a five-day sprint and live in pilot today.

Co-founder and sole engineer. Live in pilot with real employers and applicants.

Next.jsTypeScriptClaude (Haiku + Opus)Neon PostgresDrizzleClerkBulkSMS

The load-bearing details

  • 01Three apply paths with graceful degradation: if the AI budget runs out, the deterministic guided path still works, so nobody is ever unable to apply.
  • 02Fairness by construction: a whitelist strips personal identifying information before ranking, and any ranking claim without cited evidence is dropped in code rather than by asking the AI nicely.
  • 03SMS-first identity with securely hashed one-time PINs, database-backed rate limiting and delivery-status polling, so nobody is told a code was sent when it was not.
  • 04Typed translations: the build fails if any of the four languages is missing a single phrase.
  • 05POPIA (South Africa's data protection law) built in from day one: versioned purpose-specific consents, audited contact reveals, a deletion tool and a consent-derived talent pool.