Why ZoNuLy exists
The problem, the bet behind it, and the rules it never breaks.
ZoNuLy was built by final-year B.Tech (CS + AI) students at Newton School of Technology, class of 2027, who decided not to depend on campus placement and to find their own way into well-paid engineering roles.
Their work is shipped work — fine-tuning on rented GPUs, a database migration that cut infrastructure cost by 95%, a founding-team AWS build, a retrieval chatbot, a freelance marketplace with a thousand members. What they are not is competitive programmers, and the hiring processes at famous companies are built around exactly that.
The opportunity#
An engineer sitting in an office in the US, UK or Germany costs the employer ₹1.5–2 crore a year all-in. The same company can hire an equally capable engineer working from India for ₹20–50 lakh, and it is a rounding error to them.
That gap is not exploited by the household names — they are hyped, oversubscribed and gated by algorithm puzzles. It is exploited by recently funded startups, seed to Series B, which have money, urgency and small teams. And those companies mostly don't post on the big job boards: they post on their own careers page and wait.
Four problems, one pipeline#
| Problem | What it means | How ZoNuLy answers it |
|---|---|---|
| We never see most of the good jobs | The roles live on company-owned ATS pages, not on LinkedIn or Naukri. | Reads Greenhouse, Lever, Ashby, Workable and careers pages directly. Discovery |
| We can’t tell which are worth our time | “AI Engineer” can mean eight years of ML or a founding role a strong student fits. | Scores every role against the resume, with the reasons. Roles & scoring |
| Cold applications go nowhere | A portal submission is one of 800. A referral, even a weak one, puts you in a stack of ten. | Finds the engineers and founders who could refer you. People |
| Referrals don’t scale by hand | Twenty minutes per company, per person — and there is coursework. | Does the nineteen minutes; leaves you the ten seconds. Review gate |
Our judgement matters at exactly one moment — “yes, send that email to that person” — and that takes ten seconds if the other nineteen minutes are done.
What “done” looks like#
You open the app once a day. Overnight it found companies you had never heard of, with an honest grade and a sentence explaining it; found real people at the good ones with their role labelled; and wrote the emails. You read the queue, fix the lines that don't sound like you, and approve. That is ten minutes.
The arithmetic#
Volume is bounded by what a personal Gmail can send without being treated as a spammer — about 25 a day. That is why who it sends to matters far more than how many companies it finds. Per person, per month:
What it will never do#
- Auto-apply. No portal spam, no bulk applications. One person, one role, one ask.
- Send without you. Every email passes the review gate. See Guarantees.
- Invent anything. No project, number or connection that isn't in the source material.
- Touch your accounts to scrape. Your LinkedIn, X and GitHub stay yours; a ban would destroy the network referrals depend on.
- Send like spam. 25 a day, spread through the working day, at most three into one company, written like a person writes in Gmail, no tracking pixels.