About the Role As Lead AI Engineer on the platform side, you build the inference orchestration layer that sits between our product teams and their model providers. Routing, fallbacks, cost tracking, A/B testing for model swaps, and observability are all yours. Your customers are internal engineering teams, and your job is to give them a single reliable interface to every model the org uses while keeping the complexity of that routing layer off their plates. You also own the observability and eval-in-production infrastructure: standardized tracing and cost dashboards across all agentic products, and the shadow-testing infrastructure that lets us validate model swaps safely before they reach production traffic. What You’ll Build Build and operate a unified model gateway that abstracts provider complexity for product teams. Teams work with a clean interface; the platform handles routing, provider selection, and fallback logic under the hood Design and implement intelligent routing that matches each request to the right model based on task complexity, latency requirements, and cost targets. Not every call needs the same model Build resilience into the platform so provider outages, rate limits, and latency spikes are handled transparently. Agentic workflows stay up regardless of what happens upstream Own the observability layer across all AI-powered products: cost per call, latency distributions, token usage, and quality signals. Give product teams and leadership a clear view of how AI is performing and what it costs Build the infrastructure for safe model transitions: run new models alongside production, compare outputs, and roll out changes gradually with automated quality checks at every stage Drive continuous cost efficiency through caching strategies, request optimization, and per-team spend attribution so the org can scale AI usage without costs growing linearly with traffic What We’re Looking For 5 to 8 years as a backend or platform engineer, with a track record of building API gateways, middleware, or developer platform services at scale. Strong in Go or Python Experience building high-availability, low-latency distributed systems: load balancing, circuit breakers, graceful degradation, retry logic, and observability using Prometheus, Grafana, OpenTelemetry, or equivalent Solid understanding of LLM APIs and token economics. You can design routing rules based on input/output token pricing, streaming vs. batch tradeoffs, and how prompt length affects both cost and latency You think in platform terms. You know the difference between building for end users and building for engineers, and you know that internal platform quality shows up in other teams’ velocity Familiarity with LLM orchestration and observability tooling: LiteLLM, Portkey, Langfuse, LangChain, or similar. You do not need to have used all of them, but you need to understand the landscape well enough to make good choices Experience with Kubernetes and distributed systems. GPU workload scheduling or ML serving infrastructure is a meaningful bonus Why This Role is Different The platform you build becomes the backbone of every AI-powered product at Razorpay. Good infrastructure decisions here compound across every team and every workflow that ships on top of it You work on real scale from day one. The problems are concrete, the feedback loop is tight, and the impact of what you build shows up in production metrics quickly This role combines deep platform engineering with the emerging discipline of LLM infrastructure. It is a rare combination that puts you at the leading edge of how AI systems are built in production You are embedded in the decision-making, not downstream of it. You work directly with ML engineers and product teams, and your input shapes how the org approaches model selection, cost, and quality at every level You get genuine ownership over the architecture. This is a space where the right patterns are still being defined, and you have the scope to make meaningful design decisions rather than inherit them

About Razorpay

Overview Power your finance, grow your business. Razorpay is India’s first full-stack financial solutions company. We are on a mission to enhance the payment experience of over 300 million end consumers. And in doing so, we aim to enable Indian businesses - big and small - accept payments digitally with minimal effort and maximum ease. Razorpay has grown from being a payment gateway provider to a solutions-driven organization boasting of an extensive products suite to accept and disburse payments as well as raise capital and park money. In a nutshell, we fit into every nook and corner where your business touches money. #OutgrowOrdinary We identify ourselves as disruptors in the digital payments space and our vision is to power the financial ecosystem for other disruptors. Like attracts like and Razorpay actively looks to partner with established companies and startups that have either broken the glass ceiling in their industry or are set to. The Razorpay Product Suite today comprises verticals, along with Payment Gateway, like Payment Links, Payment Pages, Subscriptions, Smart Collect, Route, Razorpay Capital, RazorpayX, Payroll and Thirdwatch. Razorpay was started in 2014 by two IIT Roorkee alumni, Harshil Mathur and Shashank Kumar. Just a short few years later, Razorpay has evolved into a 800-odd strong organization with some of the best talents in the country helping some of the best companies manage their money movement seamlessly. Certified cool We are a bunch of spirited, ambitious and fun folks. And no, we’re not saying this ourselves--leading institutions have recognized Razorpay for the high trust and high-performance culture that we maintain. Our strength lies in the people we are and we go to great lengths to nurture a family of coders, designers, sellers, marketers, analysts, writers, runners, photographers, gamers, tinkerers, and above all, people who are dreamers and doers at the same time. Be a part of our exciting journey.