Product Engineer— Agent-Native Engineering

Product Engineer — Agent-Native Engineering
Truworth Wellness · Jaipur, Rajasthan
Jaipur · In-office · 1–3 years' experience
Read this part first
You will not be evaluated on code you write by hand.
You'll run parallel Claude Code workstreams against written specs. Your job is to decompose the spec, dispatch the work, catch the agent when it's wrong, and own what ships. The typing is not the job. The judgment is the job.
If you believe AI-generated code is inherently inferior, or that using agents is cheating, this role will frustrate you and you should not apply. That's not a culture-fit platitude — it's the actual daily work.
What you'd be working on
Truworth Wellness runs The Wellness Corner — a corporate health benefits platform serving 450+ clients including HSBC, EY, KPMG, BCG, SAP, Accenture, and Wipro. Real production systems, real users, real consequences. Not a demo.
This is a full-stack, end-to-end role. You'll own the whole surface — product research, product design, database architecture, back-end, front-end, QA, deployment, and SRE. We're not hiring for a slice of the pipeline; we're hiring someone who can take a problem from "what should this even be" to "it's live and I'm watching it."
Stack: Node.js, TypeScript, React, MongoDB, PostgreSQL, Redis, AWS.
What you'd actually do
Work across the full surface — product research, design, database architecture, back-end, front-end, QA, deployment, and SRE — not just one layer of it
Take a spec, break it into workstreams an agent can execute in parallel, and run them
Verify output — know what to trust, what to check by hand, what to throw away
Escalate when the agent is confidently wrong (this is the skill we care most about)
Ship. Own the thing after it ships.
Who this is for
1–3 years building real software (product company, startup, or services — we don't care which)
You already use Claude Code, Cursor, or similar — not as a toy, as how you work
You've thrown away code you wrote and didn't mourn it
You're currently somewhere that restricts, discourages, or ridicules this way of working, and you're tired of it
Strong CS fundamentals matter. Where you studied doesn't. If you're from an IIT/NIT and this describes you, great — but the filter is practice, not pedigree.
Who this is not for
People who want to be told exactly what to build
People who need to write every line to feel ownership
People who list "AI" as a skill but have never shipped with an agent
Process
Four steps, roughly ten days start to offer:
Conversation with the CPTO (45 min). You show me how you work, I show you how I work. This filters harder than anything else.
Spec-to-handoff exercise. We give you a real spec. You decompose it and run it.
Debug-with-agent. Something's broken. Find it, with your agent, while we watch.
Offer.
No leetcode. No DSA rounds. No take-home that takes a weekend.
Working here
This is an in-office role, based out of our Jaipur office. We're not hiring remote for this one — you'll be working alongside the team, in person, day to day.
You'd report to and work directly alongside the CPTO, who runs parallel agent streams on this codebase daily. You will not be the only person here who works this way, and you will not have to justify the workflow to anyone.
The company is 15+ years old and profitable, with a client base that renews. That's stability most places offering "AI-native engineering" don't have.
How to apply
Send us something you built with an agent. A repo, a CLAUDE.md, a thread, a Loom — anything that shows how you actually work. A résumé alone tells us nothing we need to know.

Work arrangement
No