Junior AI Software Engineer, AI-Assisted Product Delivery (Rotterdam, hybrid · Full-time)

About DocHeart

DocHeart is building the Operational Control Layer for Logistics.

Logistics operations are complex. Work is spread across emails, documents, operational systems and people, while exceptions and changing circumstances are part of everyday reality. DocHeart brings this together in one intelligent operational layer that understands what is happening, organizes the work and helps teams take the right action.

We combine AI, workflow automation and human decision-making to help logistics companies run their operations with more control and significantly less manual work. We are building a scalable logistics coordination platform, not a collection of customer-specific automations.

We are an early-stage startup with real customers across Europe, and we build the product while we scale. The upside: a lot of ownership, direct contact with customers and product decisions, short feedback loops, and real influence on how our technology and our company develop. It is also genuinely early: priorities shift, specs are often something you shape yourself, and you will sometimes improve on something we built fast last quarter. If you want to build the product and the company at the same time, you will love it here.

The role

We are looking for a software engineer who turns real customer problems into product that logistics teams depend on every day.

This is deliberately a combined role: you do customer implementations and you build the product. Roughly speaking, part of your time goes into getting DocHeart live and working inside a customer's operation, and part goes into turning what you learn there into platform capabilities. The two feed each other, and that is the point, not a compromise.

You will work across customer implementations and the DocHeart platform, building backend capabilities, integrations, workflow logic and document-processing functionality. A requirement may start with one customer, but your job is to understand how it should fit into the wider platform and how we avoid solving the same problem again for the next one.

We use AI coding agents extensively, and what great engineering looks like with them is still being invented. You will help define it here. With Claude Code and similar tools you will ship far more than a traditional team would, and the technical judgement you add on top is what turns that speed into something customers can rely on.

You delegate execution to an agent and keep the judgement, the architecture and the ownership yourself. That combination is the craft, and it is a new one.

What you will work on

• Python backend services and APIs
• Workflow and dossier logic
• Document ingestion, extraction and validation
• Integrations with customer systems
• Asynchronous processing and failure handling
• Reusable product capabilities emerging from customer implementations
• Testing, debugging and production issues
• Improving existing architecture as the platform scales

How we work

Own the result

You take a problem through to production. You care about the whole arc: edge cases, failure scenarios, testing, maintainability and what happens after deployment. When something breaks, you treat it as a chance to make the system permanently better.

Get the most out of AI

Coding agents are part of your normal development workflow. You are good at giving them context, breaking down problems, directing their work and reading what comes back. You have an eye for the strong abstraction, the right assumption and the simplest design that holds, and you steer the agent towards it. And you can talk anyone through the result: why the implementation is right, which assumptions it rests on, how it was tested and what you would watch after release.

See the whole system

You are fluent in APIs, databases, asynchronous systems and application architecture, so you can see where a change lands across the whole platform. You enjoy the moment two customer requirements turn out to be one capability, because that is where a single well-built feature starts paying off for every customer after it. And you have the instinct for which small change deserves a second look before it goes out.

Bring a growth mindset

The tooling, the product and the craft are all moving fast, and that is the fun part. You look for feedback rather than wait for it, pick up unfamiliar domains and systems quickly, and treat every customer implementation as a chance to get sharper. We will invest in your growth, and we expect you to make the team better as you go.

Ship fast, ship proud

We love engineers who move quickly and get working software in front of real users. You have a feel for which shortcuts buy real speed, you build things you are happy to put your name on, and you get a kick out of seeing your work used in a live operation the same week. Here, momentum and quality are the same skill.

Your background

• 3–7 years of professional software engineering experience
• Strong Python and backend development experience
• Experience building APIs and integrations
• Experience owning software in production
• Cloud and automated testing experience
• Experience in a SaaS, platform or startup environment
• Serious day-to-day experience with Claude Code, Codex or comparable tools

Logistics experience is useful but not required. Most of our engineers learned the domain on the job, straight from the operations teams that use DocHeart every day.

You will love this role if

• You treat AI-generated code as a strong first draft you make excellent.
• You would rather shape an open problem than wait for a detailed ticket.
• You measure yourself on what works in production, not on volume.
• You stay with a change until it is live and working for real users.
• You see a production issue as the fastest learning opportunity you will get that week.

How we assess

We would rather talk about real work than test you on puzzles. Tell us about:

• Something you built and shipped.
• A production problem you personally owned.
• How you use AI agents in your development workflow.
• A time you overruled your agent, and how you know when AI-assisted work is genuinely ready to go live.

No algorithm puzzles. We care about engineering judgement, ownership, curiosity and the quality of what you build.

Sound like your kind of build?

Send us the thing you are proudest of shipping and a few lines on how you work. We read every application and we reply.

• Please note: this is a hybrid role based in our Rotterdam office. We do not offer relocation support or visa sponsorship, and we do not work with recruitment agencies.