Senior Product Engineer (Backend Leaning) in Berlin

• Senior Product Engineer (Full Stack) — AI-Native B2B Trade Platform, Berlin (Hybrid)
• Via Xcede — confidential search
• The company

A fast-growing, well-funded Berlin-based startup rebuilding B2B procurement from the ground up.

This category is a €900B+ market in Europe alone, still run almost entirely by people — four or five intermediaries between supplier and customer, with up to 40% of the margin evaporating along the way.

This company is building the alternative: they buy from global suppliers and sell directly to European customers on their own account, running sourcing, pricing, sales, logistics, credit, and financing through AI.

One request in, a firm quote back in minutes, delivered on time — no warehouses, no fleet, no inventory risk, every trade pre-sold before it’s bought.

They’re also the credit counterparty, scoring and insuring customers in minutes with working capital that’s bank-funded, not equity-funded.

Where they are today: strong triple-digit-million-euro turnover run rate, majority repeat customers, seven live European markets sourcing globally, 80+ people, and a recent well-backed funding round from top-tier European VCs plus bank working-capital facilities.

The role

Much of this company’s trade execution runs through AI agents — and agents are only as good as the product around them: reliable data, well-designed APIs, and interfaces where the team can review, correct, and approve what an agent proposes before real money and real shipments move.

As a Senior Product Engineer, you’ll own business problems end to end — talking to the people who use what you build, deciding with Product what to solve, and shipping the whole slice: data model, services, AI agents, and UI.

You’ll work in a domain where prices and stock change daily, supplier data arrives in every format imaginable, and a wrong number costs real money.

• What you’ll do
• Own a business problem from first conversation through production — e. g. making supplier pricing reliable from ingestion all the way to the quote a salesperson sends.
• Build backend services for pricing, inventory, orders, and integrations across suppliers, customers, and logistics partners.
• Build the AI agents themselves plus their tools and APIs, with safeguards — idempotency, validation, audit trails, human review where it matters.
• Design the interfaces where commercial teams review, correct, and approve agent proposals.
• Turn messy external data (price lists, datasheets, stock feeds) into data the business can trust.
• Evaluate and monitor LLM-driven workflows in production.
• Lead technical decisions, document trade-offs, and mentor other engineers.
• Help shape hiring — run interviews, help define what good engineering looks like.
• Tech stack

Java 21 / Spring Boot 3, Python, Lang Graph and LLM APIs (Open AI, Anthropic, Gemini), Postgre SQL, Kinesis, AWS + Terraform, Git Hub Actions, Type Script / Node. js / Next. js / React.

You don’t need all of it on day one — the team leans heavily on AI coding tools and expects you to pick up whatever a given problem needs.

What you bring

• 5+ years building and running production software with real users and real consequences.
• Strong backend skills at production scale in Java or a comparable language (Kotlin, C#, Go) — willingness to work in Java/Spring Boot specifically.
• Experience designing distributed, event-driven systems with a focus on data correctness and failure modes.
• Solid Postgre SQL and hands‑on AWS experience.
• Full‑stack delivery capability — your depth is backend, but you can ship a good interface in Type Script/React when the problem needs it.
• Strong product sense — you ask why before how, talk to users, and turn vague problems into clear plans.
• Comfort in a fast‑growing, fast‑changing environment.
• A track record of solving business problems, not just technical ones.
• Curiosity and good engineering judgment about LLMs/agents is a must; hands‑on experience building with them is a strong plus.
• Bonus: experience in trading, marketplaces, supply chain, fintech, or another domain where data errors cost real money.
• Practical details

Berlin, hybrid. English‑speaking environment. Learning/development budget included.