Senior Software Engineer I/II - Backend (Python)

About UrbanPiper

Close to 30 million restaurant orders a month run through UrbanPiper. We sit between thousands of restaurant kitchens and everywhere they sell: Swiggy, Zomato, their own storefronts, the POS at the counter. When we get it wrong, a restaurant stops selling. When we get it right, nobody notices. That's the job.

About the Team

A restaurant's menu is its storefront, and it lives in a dozen places at once. Change one price, one add-on, one opening hour, and it has to land correctly on every aggregator, each with its own schema, its own rate limits and its own ways of failing silently. Now do that for thousands of brands, hundreds of times a day at peak.

The Menu team owns that pipeline: ingestion, publishing, and knowing exactly what went where when something breaks. The hard parts are : data modelling, idempotent sync, async processing at scale, and partners who change their APIs without telling you.

How we work

Small team, large surface area. We use AI as a force multiplier & not just to speed up coding. We expect you to work effectively with AI agents, exercise sound engineering judgment, and take ownership of what ships. Not every problem comes with a clear spec or an obvious solution; we expect you to iterate, challenge your assumptions, and find the best solution to the problem.

What you'll do

• Ship to production in your first week. Then keep shipping.
• Take a merchant's "my prices are wrong on Zomato" from Support, find the cause in the logs and the data, and make sure it can't happen again.
• Design async pipelines that survive retries, partial failures and rate limits (Kafka, RabbitMQ, SQS, Temporal).
• Model messy catalogue data so that each new aggregator's quirks stay out of the core.
• Build the validation loops that make AI-written code safe to ship: tests, replays against real production payloads, staging checks.
• Instrument your services well enough that you find the bug before the merchant does.
• Review code, write design docs, and go on call for what you built.

You'll fit if

• You're AI-native. You run AI coding agents (Claude Code, Codex, Cursor or similar) every day, and not as ask-and-hope. You think before you prompt, give the agent the right context, and close every loop with a check. You have opinions on models, tools, plugins and skills, and you can defend them.
• You take responsibility for what ships, regardless of who wrote it. AI makes mistakes; you know where, and you catch them before production does. You own the code whether you typed it or not.
• You've built Python backends that run in production at meaningful scale. Django or FastAPI (we run both), REST and GraphQL APIs with versioning, pagination and error contracts.
• You know your fundamentals. Data structures, testing, SQL you can index and read a query plan, and the ways async systems fail.
• You write clearly. A PR description or incident note from you explains the trade-off.
• You've owned something. A service in production, a design across teams, engineers who got better working with you.

Good to have

• You have built your own AI tooling: agent skills, MCP servers, hooks, eval sets, or automations that run without you.
• Hands-on experience with Kubernetes, Helm, and GitOps with ArgoCD.
• Catalogue, e-commerce, or third-party marketplace integrations.
• Durable workflow engines such as Temporal.
• Experience at a fast-paced startup.

Your first 6 months

• Week 1: ship your first PR to Prod! You set up your environment and AI tooling yourself, from the code and the docs.
• Month 1: you take a real customer escalation, root-cause it from logs and data, and ship the permanent fix. You're on the on-call rotation.
• Month 3: a part of the menu pipeline is yours, end to end. Something you shipped is used daily by partners or Support, with alerts and dashboards you built.
• Month 6: you own a number, such as publish failure rate or ingestion time, and you've moved it. You've led a design across services, contributed to next quarter's roadmap for your area, and built tooling that makes the rest of the team faster.