AI Product Manager
Anyone can build anything now. Almost nobody knows what to build — and even fewer
can take it to market.
Read this before you read anything else
We are extreme about four things, and nothing else:
Extreme hard work. Extreme respect. Extreme shipping. Extreme innovation.Not hours. Nobody here counts your hours. We count what you shipped and what it did inthe real world. If you need to be told when to start, when to stop, or what to do next — close thisletter. You'll be unhappy here, and so will we. This is not a boss-and-employee relationship. There is no boss. There's a product, a frontier, and you. You'll be rewarded and awarded on exactly one thing: your hunger. Not your title, not your years, not who you know. The one word Hunger. Then grit. Then go-to-market. Everyone can come up with ideas — ideas are the cheapest thing in this building now . Go-to- market is the whole game. Taking a raw idea and thinking it end-to-end: the user's real pain, theempathy, the pricing, the packaging, the last mile into production and into a customer's hands. Taking it there — that's the job. That's what we hire and reward. This is not a place where tasks get handed to you. This is where you create your ownresponsibilities. This is where you build tomorrow's innovation — before anyone asks you to.
The 90 / 10 (HAT – Humans And Tokens)
Here's the philosophy this whole place runs on. AI shipping is in your hands — the PM's — not someone else's. You don't request it. You do it. 90% of the work is done by the tokens. 10% is done by you. Your Claude Code session does the 90% — and yes, that means you actually code. You ship things to a very high perfection straight into the developer's hands; the developer takes a look and takes it to production. That's how it's wired here. But the 10% is the whole reason you exist — and it's the hard part: the design standard
the user empathy
the elegance
the choice of which route to take
the delight
That 10% of human thinking is the real craft. It's what I'm actually hiring. The machine can't do
it. You can. Choosing is the job
AI made building cheap. Claude Code turns an idea into a working prototype in an afternoon —so the old PM job (write the spec, wait for engineering, ship next quarter) is dead. We buried it. RIP the 40-page PRD, you will not be missed. What's left is the hard part: choosing, then shipping. Which agent behavior makes a CMO lean forward, and which is a demo that dies in production? Where's the line between an insight and a hallucination wearing a suit?
Nobody here has written a PRD in years. You build it — with your own hands, in Claude Code — and let it argue for itself. When it wins the argument, you drive the code so close to deployable that dev's job is a double-check, not a rebuild — and it ships that week. Showing beats telling. Shipping beats showing. Every single time. But don't mistake speed for vibes . Every bet gets written down first — one page, we call it a P-B-S-E: Problem — who's hurting, what are they doing about it today, and why now? Bet — a falsifiable hypothesis: if we shift this, the user will observably do that, within this much time.
Success Criteria — the concrete behaviors we'll see if the bet is right.
Evaluation — how we measure it, the threshold that kills the bet, and the threshold that
scales it.
Then you build. The PRD died here; the discipline didn't.
The loop
We are always on. A technique drops on Twitter Tuesday morning; by afternoon someone here has tried it; by Thursday, if it's real, it's in the product; by Friday a brand team on the other side of the world is using it. See it. Build it. Ship it. The loop never stops, because the frontier never stops. If your learning lives in "read later" folders , this will break you. If you're the person whose friends ask "how did you already know about that?" — you're already living the loop. You just don't get paid properly for it yet.
The company you'll keep
Our product managers are the smartest in this industry — I say that as a warning about the bar, not a compliment. They don't bring ideas to meetings; they bring code close enough to done that engineering runs on their branches. Owner mindset as the entry requirement — they own the product, the client's Tuesday, the number. They drink AI credits like water — nobody meters that here; we'd rather fund a thousand of your experiments than one status meeting. Everything happens in the open: no closed rooms, no politics to decode, best argument wins regardless of whose it is. If you've spent years learning to manage up , you can stop. That energy goes into the product now. They'll make you better or make you tired. Ideally both.
Your first week
Day one, morning: you're inside ARIA — our AI agent that watches millions of social videos in Hindi, Tamil, Bahasa, Thai, and turns them into intelligence for the world's biggest consumer brands . Real system, real traces, real client questions in the queue. Day one, afternoon: you pick one. By the end of the week: you ship the answer. Not a deck. Not a prototype for engineering to "productionize later." The thing itself, live, built in Claude Code, in front of that brand team's eyes.
If that scares you, this isn't your room. If you're already thinking about how you'd build it —
keep reading.
What you own
The product's judgment. What ARIA does next, what it refuses to do, the evals that define"good." Find a problem and you don't fix the instance — you build the eval that kills the whole class. Your choices reach production carrying your commits. Your ideas never sit in someone else's backlog. And the room. This role is client-facing to the bone: you'll sit with brand teams at the biggest consumer companies on the planet, get inside their psychology until you can predict their next objection, and use the product exactly the way they do. We build products with a GTM mindset — sharp enough, self-evident enough, that they do their own go-to-market. Every feature you ship should sell the next meeting. No product person between you and the user. You are the product person. The bar is craft, not years. I don't care what your last title was. I care what you've shipped — and
what you chose not to. Who this is actually for
This is a close-to-100% self-starter environment. You do not want to be here if you want to be led. You want to be here if you believe you are the only one who can take this product where it needs to go — the king or queen of it — and there's absolute hunger burning in you to do it. Here, will is bigger than skill. And AI isn't atool you use; it's core to how you think.
Fair warning
The pace is relentless and the problems don't have playbooks — the industry is still writingthink-pieces about agents while we run them in production for enterprise clients. You'll be wrongin front of smart people, often , and expected to be less wrong by Friday. If you needcertainty before you build, you'll be miserable here. If you build to get certainty, this is the best room you'll ever work in — what you learn here in one year, you won't learn anywhere else in five.
How this goes
Don't send a resume. Send a choice — something you built because you decided it should exist, with a line on why you built that and not the ten other things you could have. The choosing iswhat I'm hiring. (The top AI labs noticed something last year: nearly every product person they hired had a GitHub. Same bar here. It doesn't need stars — it needs to exist.)
Pay: ₹2,000,000.00 - ₹7,000,000.00 per year
Application Question(s):
- Did you graduate from an IIT(Indian Institute of Technology)? If yes, please specify which campus.
Education:
Experience:
- Product management: 3 years (Preferred)
- AI Product Manager: 3 years (Preferred)
- Product lifecycle management: 3 years (Preferred)
Location:
Work Location: Remote
