About the role
The opportunity
Razorpay is building the platform where AI agents do real work for businesses. Agents act on live business events, reach into a company's systems through integrations, talk to customers over chat and voice, and show the value they delivered. Businesses can switch on ready-made agents, or describe a problem in plain language and have an agent built, tested and deployed for them.
The product is live and growing quickly. The next stage is the hard one: turning an AI-first product into a platform that stays reliable at scale, costs less to run as it grows, is safe by default, and lets other people build on it with confidence.
The role:
You will lead the engineering team at the centre of this work. The problems combine distributed systems, applied AI and product design, and they shift as fast as the models underneath them. Few answers are settled, and many decisions have to be made with partial information.
Your impact goes beyond what your own engineers ship. Product managers, designers, engineers and business teams all build on or depend on this platform. You will make each of them more effective: clearer priorities, better tools, faster feedback, and an AI-first way of working that you practise yourself.
What you will do:
- Scale an AI-first platform: Own the architecture and the path from early traction to production scale. That includes event-driven execution, multi-tenant isolation, capacity planning, and staying resilient when model providers and third-party services hit their limits. Set reliability targets and run on-call and incident reviews that leave the system stronger each time.
- Make quality and cost measurable: Build the evaluation practice that decides when an agent is good enough to ship and whether it keeps getting better. Own model choice, routing and unit economics, so the platform improves and gets cheaper as models evolve.
- Build safety in: Agents here act on payments and customer data. Make isolation, guardrails, human approval for high-risk actions and auditability the default. Work with security, legal and compliance on what enterprise customers need.
- Empower everyone who builds: Make the platform something product managers and designers can prototype and ship on, not just engineers. Raise the bar for AI-native development with coding agents, evals, instant previews and short feedback loops, so a small team moves like a much larger one.
- Align many stakeholders: Work across product, design, data, security, infrastructure, sales and customer teams. Turn competing requests into one roadmap with clear owners. Explain trade-offs plainly to engineers and executives alike.
- Grow the team: Hire, coach and keep strong engineers across backend, applied AI and frontend. Draw clear ownership lines and spread architectural knowledge, so the team never depends on one person.
What success looks like in your first six months:
- Everyone knows the top priorities, who owns each one, and why.
- The platform has an agreed plan for production scale, with reliability and cost targets that are measured every week.
- Product managers and designers ship meaningful changes on the platform with less hand-holding from engineering.
- Engineers spend less time firefighting and more time on work that compounds.
- The team is growing, with a hiring plan under way and ownership that no longer sits with one or two people.
What you bring:
- You are AI-first in how you work, not just in what you build. You already use AI across the whole engineering loop: designing, writing and reviewing code, testing, shipping, running on-call and handling incidents. An AI-first platform needs a leader who has already lived in the agentic way of building software and can bring a team along.
- 8+ years building and running production systems, including 2+ years leading engineering teams.
- You have built or scaled a platform that lets many more teams build on top of it, faster and with less help: shared foundations, clear contracts and self-serve paths that multiply what others can ship.
- Depth in distributed systems at scale: reliability, multi-tenancy, capacity, rate limits and cost.
- You have shipped LLM-powered products or agents to production, hands-on with agent frameworks, LLM gateways and LLM tracing or observability, and dealt with what follows: evals, model changes, cost and guardrails.
- Sound judgement under ambiguity. You make calls with partial information and change course in the open when the facts change.
- You communicate clearly with engineering, product, design and business audiences, and people leave your meetings with less confusion than they arrived with.
Nice to have:
- Voice AI, conversational messaging or customer-communication platforms.
- Payments, fintech or another regulated domain.
- Products where non-engineers create software, such as no-code, low-code or AI builders.
Why this role:
Few roles combine a live AI product with real customers, a platform other teams build on, and the chance to shape how a whole product organisation works with AI. You will do all three, with a team that already moves fast and wants a leader who helps it move further.