Zonko is a frontier AI lab building the first truly global, daily-use product born out of India. We ship fast, watch what spreads, and compound taste until something breaks through.

This page is about whether you should actually work here.

Probably not.

Why you might not want this

We don't track hours. We track artifacts.

Every single day should produce something — shipped, documented, or learned. If you need "reasonable" workloads, this isn't it.

Nobody tells you what to do.

You own your stream. You get stuck, you find the answer — papers, chatgpt, claude, github, other developers, cold emails, etc. We're building from zero, not running a playbook.

Your ideas will die fast.

We close threads quickly. If something doesn't work after a real attempt, we move on. No emotional attachment to your code.

There's no small talk.

When Cursor's cofounder was asked what worries him about culture, he said: "People start talking about the weather at meals." That's our nightmare.

You'll be in office.

In Mumbai. Every day. Remote isn't an option. The best work happens when high-agency people collide in the same room.

You'll fail constantly.

We run experiments. Most don't work. If failures make you doubt yourself instead of sharpen your intuition, you'll burn out.

We don't care about credentials.

IIT, Stanford — irrelevant. Can you build? Have you shipped? Do you have weird hobbies and builder energy?

We go deep

Not surface-level "prompt engineering" deep. Actually deep. Reading papers. Understanding architectures from first principles. Knowing why something works, not just that it works.

When a new model drops, we don't just try it — we dig into the technical report, understand the training choices, figure out what it means for what we're building.

When something breaks, we don't just patch it — we understand the system well enough to know why it broke and how to make sure it never breaks that way again.

We want people who nerd out on the details. Who find joy in understanding how things actually work under the hood. Who can hold a conversation about attention mechanisms and distributed systems and product intuition — all in the same breath.

Open roles

AI Engineer

1-4 years · Mumbai · ₹40-90 LPA

You will:

  • Build AI products end-to-end — from prototype to production
  • Work across memory systems, agents, tool use, and AI infra
  • Read papers and implement ideas the same week
  • Own entire streams, not just tickets

You might be a fit if:

  • You have shipped side projects — things that actually work, not just demos
  • You understand harnesses, prompt engineering, evals, and why they matter
  • You can hold your own in a conversation about agents, tool use, and ai infra
  • You don't need a spec to start. You are the spec.

Backend & Infra Engineer

1-4 years · Mumbai · ₹40-90 LPA

You will:

  • Build and operate systems that scale — latency, reliability, uptime
  • Own infrastructure decisions from day one
  • Make tradeoffs between moving fast and being paranoid
  • Debug production issues at 3am (and make sure they don't happen again)

You might be a fit if:

  • You've built and operated production systems before
  • You think about edge cases before they happen
  • You love the plumbing — not just the "AI part"
  • You find satisfaction in systems that just work

How to apply

Resume optional. Proof of work mandatory.

Send links to things you've built. GitHub, demos, shipped products — anything real.

One paragraph on why this. Not "why Zonko." Why this mission? Why AI products that reach billions? Why now?

Something weird about you. A hobby, an obsession, a rabbit hole. Anything that shows you don't think like the median.