Product · Mumbai (on-site) · Full-time

Product Engineer

Live at the intersection of product, design, and engineering: take a feature from "this just landed in my lap" to deployed, and decide for yourself what's worth building.

Cityflo moves 60,000+ commuters a day, and the small decisions (what to surface, what to ignore, what to ship today versus next week) are the product. You'll own features end to end: ideation, design, frontend, backend, deploy. You'll work next to product and design, but you'll often be the one deciding who the user is and what the one concrete problem is. We lean on AI heavily; you'll need taste for when it's right and when it's confidently wrong. This is for someone early in their career who already ships independently and thinks user-first by reflex.

What we look for

• 1–4 years building real product: you've shipped features end to end, not just closed tickets
• Comfortable across the stack: React on the front, Python (Django) or Go on the back, REST, some SQL/NoSQL, Git, CI/CD, an AWS deploy
• Strong product instinct: you reach for the user and the problem before the framework, with a real feel for UX, accessibility, and responsive design
• You use LLMs in production work: prompt engineering, API integrations (OpenAI/Gemini/etc.), and a sense for where AI actually earns its place
• You take ownership and deliver independently: you make the call, write down why, and move
• You treat user input and data as untrusted by default, and you verify what a model tells you instead of trusting it

The assignment

A spreadsheet of ~30 recent rider feedback and support messages just landed in your lap. You're the on-call product engineer; nobody has triaged it, and standup is soon. Write a short memo (who's the one user, the one concrete problem, what you're deliberately cutting), then build a small web view that uses an LLM where it actually matters to surface the 2–3 things that genuinely need attention (triage / cluster / draft a reply). A real UI, not a notebook or a CLI. Aim for 2–4 focused hours. Driving your coding agent is expected; that's the job.

Aim for 2–4 focused hours. This is intentionally small: a sharp, well-reasoned slice beats a sprawling one, and going much past 4 hours is a signal we'd rather not see. Lean on your agent fully; that's how we work.