Are you looking to break out of the silo of pure-tech engineering and build systems that directly move millions in global revenue? If you want to own an entire, high-impact platform end-to-end and see the immediate business value of your code every single day, this is your playground. At Ultrahuman, you will act as a critical bridge between deep engineering and global market execution, turning data intelligence into autonomous, revenue-driving action across 100+ countries.

About the Company:

Ultrahuman is one of the world's fastest-growing wearable companies. The Ultrahuman Ring AIR the world's lightest sleep-tracking smart ring sells in 100+ countries. We operate like a product company in every function, and marketplaces is no exception: our Amazon business across the UK, Germany, Canada, Australia, India, the UAE and beyond runs on uh-amazon, an in-house platform we built ourselves.

Role: Marketplace Platform Engineer

Location: Begur, Bengaluru

Schedule: Hybrid.

Marketplace Platform Engineer

Turn a read-only intelligence platform into the hands that run a global Amazon business.

Why this role exists:

uh-amazon already does the seeing: a fleet of agents producing daily intelligence across every marketplace, an automated fulfillment middleware that intercepts and routes every order, and a self-hosted agent platform the team uses without writing code. But it writes almost nothing back to Amazon every ads change, price correction, and listing push is still done by hand. That write side is unbuilt, and it is worth more than the entire team's cost: the daily briefs already flag large sums of recoverable ad spend that no one has the hands to recover. You will be the platform's owner and the team's force multiplier.

What you'll own:

- The uh-amazon codebase end-to-end the FastAPI middleware, the agent fleet, the self-hosted agent platform, and the deploy pipeline

- The write side, behind staged approval gates the order-management bridge first, then ads execution, repricing, and listing-push agents, each earning the right to act

- Platform trust kill the false-positive classes and fix the chronically failing agents, so the team acts on the briefs instead of second-guessing them

- Operator self-serve the team adds and tunes agents through the platform without touching code

Your first 90 days:

- Ship the order-management bridge that removes the last manual step in fulfillment

- Ship the first ads-execution agent behind a human approval gate, working the platform's existing recommendation lists

- Eliminate the top false-positive classes in the daily briefs

What we're looking for:

- Strong Python (FastAPI or similar) and real API-integration depth OAuth, rate limits, retries, idempotency; you have built against messy third-party APIs in production

- LLM-agent experience you have built tool-using agents or LLM pipelines, designed prompts and tool schemas, and debugged why a model did something

- Ship-fast evidence small PRs, production ownership, comfort being the only engineer on a system

- Comfort owning production writes against a live revenue channel, with gates and audit trails you design

Strongly preferred:

- Amazon SP-API or Ads API experience specifically this collapses your ramp to near zero

- E-commerce domain (order management, marketplaces, repricing, feeds); container/orchestration experience

- A public GitHub with real contributions