Medior AI Product Engineer (Full-Stack

Amsterdam - Hybrid (minimum 50% at the office)

36-40 hours per week

€3,800–€4,800 gross per month, based on 40 hours and depending on experience

Futurewhiz is a B Corp

Generative AI can create textbook content in seconds. Producing content that is accurate, curriculum-aligned, editable and trusted by authors and editors—and doing so reliably at scale is much harder. That is the challenge you will help us solve.

Futurewhiz is looking for a Medior AI Product Engineer to build practical AI-powered tools for educational content creation. You will turn promising experiments into dependable software that people can use in real publishing workflows. This is primarily a software engineering role, but not one where you wait for fully specified tickets: you will help investigate problems, shape solutions, test ideas and decide what is worth building.

You do not need to be an expert in every part of the stack. You do need enough practical experience to own and ship software with a good degree of independence, plus the curiosity to use AI tools thoughtfully when working beyond the edges of your existing expertise.

For nearly 15 years, Futurewhiz has been the market leader in EdTech in the Netherlands. Our brands – including Squla and Study Go – help children and teenagers aged 3 to 18 learn and grow at their own level and pace. Our mission is clear: "Empowering every child to grow." We’ve been a certified B Corp since 2020, which means we strive for a strong balance between purpose and profit.

You will join the Futurewhiz Content Automation team and work closely with software engineers, editorial specialists and colleagues at a major educational publisher. Together, you will improve the workflows used to create and maintain high-quality learning materials, including textbooks and digital content.

Each cross-functional team focuses on a particular part of the publishing workflow. The team studies where time, quality or cost is being lost; tests possible improvements with users; and builds the solutions that show the most promise. Once the best opportunities in one area have been addressed, the team may move to another part of the workflow.

Your editorial colleague will lead much of the specialist research, such as interviewing authors and editors, testing editorial concepts and mapping publishing workflows. You will participate where technical understanding or rapid prototyping is valuable, but your primary responsibility is to design, build and operate the software.

Specifically, you will:

• Build and own AI-assisted applications from early prototype through to reliable production use

• Develop across the stack:
  Python services and APIs, data flows, user-facing interfaces and the infrastructure needed to run them

• Build and improve LLM and agent-based workflows, including prompts, evaluations, feedback loops and safeguards

• Turn findings from editorial research and user testing into practical technical approaches

• Create small experiments to resolve uncertainty before committing to larger builds

• Integrate new tools into existing publishing systems and contribute reusable components to our content-automation platform

• Use AI coding tools as part of your everyday work while reviewing, testing and taking responsibility for the resulting software

• Deploy, monitor and troubleshoot what you build, balancing quality, reliability, speed and cost

• Contribute to architecture, CI/CD, security, code quality and technical decision-making within the team

• Travel to Groningen for one or two days approximately once every two months to work with colleagues there

There is no separate team waiting to take over UX, infrastructure or operations. You will have support from experienced colleagues, but the product team owns its solutions end to end and deliberately keeps them as simple as possible.

In your first three months, we would expect you to:

• Understand the users, workflow and technical environment in your initial problem area

• Ship a useful prototype or product improvement and evaluate it with real users

• Take ownership of a defined part of the product, including its deployment…