AI Product Manager

Compensation: ₹12L – ₹20L • No equity

You will own AI-native product surfaces end to end - from problem framing through model behaviour, evals, UX, working code and adoption on the hotel floor. This is not a roadmap-admin role. You will scope the feature, write the PRD, tune the prompt, then build it yourself in Claude Code, Cursor or Antigravity and ship the PR to engineering for review.

What you'll do

• Own AI product behaviour
• Define how our agents should behave - intent capture, tool use, handoffs, confirmations, refusals, escalation to humans.
• Write and iterate system prompts, knowledge structures and conversation flows with engineering, not just requirements for them.
• Decide when a problem needs an LLM, a rule, a form or a button - and push back when AI is the wrong answer.
• Scope it, build it, ship the PR
• After scoping, build the feature yourself with Claude Code, Cursor or Antigravity - working code in our repo, not a mock.
• Open PRs with tests and a clear description; engineering reviews, hardens and merges.
• Own the scope-to-PR loop for small and mid-size features; hand off to engineering where architecture, scale or security calls for it.
• Prototype prompt, flow and UI changes against the real stack to validate before they take roadmap time.
• Follow codebase conventions, CI checks and review feedback - PRs that land, not PRs that get rewritten.
• Build and run evals
• Set up eval sets, rubrics and golden conversations for chat and voice agents.
• Catch score inflation and false confidence - make sure the numbers reflect what guests and staff actually experience.
• Turn transcript reviews and failure analysis into prioritised fixes and roadmap items.
• Ship across guest and staff surfaces
• Write clear, PM-level PRDs and TRD reviews for features across products.
• Own problems end to end, from concept to delivery & monitoring.

What you bring

• 4+ years of product management in B2B SaaS, enterprise software or platform products.
• Hands-on experience shipping at least one LLM-powered product or feature to real users - chat, agents, voice, search or copilots.
• Daily user of AI coding tools - Claude Code, Cursor, Antigravity or similar - with code you wrote merged into a production codebase.
• Working fluency with how LLMs behave: prompting, RAG, tool/function calling, structured outputs, hallucination modes, context limits.
• Writes sharp documents. PRDs that engineers can build from and leadership can decide from.
• Strong product taste for interaction cost - you notice an extra tap, an extra turn, an extra confirmation.
• Owns outcomes in an early-stage, high-ambiguity environment without needing a fully formed process.

Nice to have

• Hospitality, travel or luxury retail experience - hotel ops, PMS, guest experience or booking.
• Design sensibility - comfortable in Figma, prototyping flows yourself.