Forward Deployed Engineer

Talonic GmbH; Berlin; Vor Ort; EUR 80.000 - 90.000; Vollzeit

Benefits dieser Stelle

Equity ownership

Holiday 30 days

Hybrid work Berlin

Zusammenfassung

Talonic GmbH in Berlin is seeking a highly technical professional to design data models, schemas and extraction pipelines on the Talonic platform. You will sit in customer workshops and translate messy contracts, invoices and diagrams into a coherent data model that can be queried.

You will own delivery end-to-end, build the downstream layer, and create agents and MCP integrations that let customers run on SAP, Ivalua, Dynamics or Snowflake.

Qualifikationen

• You design deep data models for messy enterprise data.
• You build schemas and extraction pipelines on the Talonic platform.
• You own delivery end to end for named accounts.
• You communicate technical decisions to stakeholders.
• Fluency in English; German is a strong plus.

Aufgaben

• Sit in customer workshops and design the data model.
• Design and implement schemas and extraction pipelines.
• Build downstream integration with SAP, Ivalua, Dynamics, Snowflake.
• Prototype with AI tooling and harden for production.

Kenntnisse

Data modelling

Backend development

ML / AI tooling

Product instinct

English fluent

German language

Tools

AI tooling

Jobbeschreibung

We are 13 people. We run Talonic in production for energy utilities, pharma distribution, automotive and US logistics. The platform works. What it takes to win each account is someone who can go deep on a real, messy customer problem and come back with a system that runs.

You are not handed tickets. You are handed a problem, a customer, and the platform — and you are expected to come back with something that works. We care much more about what you can ship than what your CV says you are. Our strongest builder came to us as a product manager.

What you’ll do

The charter

Sit in customer workshops and leave with the data model in your head

Understand a messy real-world process — contracts, invoices, scanned engineering diagrams, forty-year-old archives — and design the schema that makes it queryable.

Design and build schemas and extraction pipelines on the Talonic platform

From data model to working extraction. You design it and you build it — not separate people.

Build the downstream layer

Validation rules, normalisation, delivery into SAP, Ivalua, Dynamics, Snowflake — whatever the customer runs. The integration is part of the job.

Build agents and MCP integrations that put customer data to work

Prototype fast with AI tooling, then harden what survives contact with production. The prototype is not the delivery.

Own delivery end to end and be the technical face of Talonic to the account

Enterprise stakeholders, procurement, IT security, a skeptical CFO. You can hold the room and you can explain the decision they are paying for.

Who you are

The profile

// Strong fit

• You build constantly. You have shipped things nobody asked you to build. The instinct to make something is more important to us than the title on your CV. Show us what you’ve built.
• You are fluent with agentic coding tools and you use them as leverage, not as a crutch You can read the code they write and you know when it is wrong. AI is how you move faster — not how you avoid understanding what you ship.
• You have real technical depth somewhere Data modelling, backend, systems, ML, industrial process engineering. We do not care which one — we care that it is deep enough to matter when things get hard.
• You have product instinct You can tell the difference between what a customer asked for and what they need. That distinction is the job.
• You can hold a room Enterprise stakeholders, procurement, IT security, a skeptical CFO. You have explained a technical decision to someone paying for it and they trusted you.
• You dive into a problem until it is finished Reliability matters more here than speed. Prototypes that get declared victories are not deliveries.
• English fluent. German is a strong plus for our DACH accounts. Most of our enterprise customers are in the DACH market.

// This isn't the role if

• You want a defined backlog and a clear spec The spec is the customer problem. You write the backlog.
• You want to work only on core platform internals and never see a customer Customer delivery is not a side-effect of this role. It is the role.
• You ship a prototype, declare victory, and move on Production hardening is the job. The prototype is the starting point.
• You have never had to explain a technical decision to someone who is paying for it That moment is part of every delivery here. It is not optional.

The operating model

Small team, high trust, context over control. Real ownership from week one, and a weekly heartbeat everyone ships to.

Monday

Monday Business requirements set

Thursday

Thursday Delivery

Friday

We are genuinely excited about where this technology is going and we follow it because we want to. You will be expected to do the same — not because it is your job, but because that is the kind of people who work here.

We help write the rules of this category

• We co-authored DIN SPEC 91491, Europe’s first standard for AI-generated data structures, with Fraunhofer IIS, Humboldt-Innovation and DIN e.V. The standard is the credibility layer — in regulated enterprise, it opens doors.
• NVIDIA Inception member, with roots in research affiliated with Humboldt University of Berlin, and backed by leading European deep-tech investors.
• Production in regulated enterprise — energy utilities, pharma distribution, automotive, US logistics. Real problems, real customers, real data. You own named accounts from month one.

What we offer

• Location Berlin, hybrid.
• €80,000–€90,000 calibrated to experience.
• Equity Meaningful early-stage ownership.
• Holiday 30 days.
• Scope Direct ownership of named enterprise accounts from month one. No ticket queue. No handoffs.

How we’ll meet

The process

No take-home algorithm puzzles. No five-round loop.

30-minute call with Nikolas (CEO) or Holger (CTO). The role, the problem space, your questions. Come ready to talk about something you have built and the decisions you made.

Paid work sample. We agree on something real. You decide how to build it. You walk us through what you made and why you made it that way — not what the spec asked for, what you chose and why.