Product Manager AI Search (m/f/d)

moebel.de is one of the leading online shopping portals for furniture and furnishings in Europe. Our new project Piqit is an AI-powered shopping assistant that is changing the way people find and buy products online.

• In the thick of it from day one: You will be responsible for the answer quality of Piqit, our AI-powered shopping assistant, which finds the right product from over 100 million offers – from initial observation to rolled-out changes.
• Quality here doesn't mean number of hits, but answer: You will check if Piqit has understood what it's about, if the suggested products are really the right ones, and if the categorization contributes to this.
• Root cause analysis is your core discipline: If an answer is not convincing, you will confidently distinguish whether suitable products were not found at all, whether an attribute is missing or incorrectly assigned in the product data, or whether the model has made something wrong from good data.
• You know the peculiarities of language models: strong in intent, nuance, and language, unreliable with exact numbers, completeness, and up-to-dateness – and always sounding convincing. You bring exactly this intuition to every decision.
• Hands-on instead of concept paper: You will write and roll out prompt sets, hints, and rules yourself, refine attributes and facets yourself, build test sets yourself – and automate recurring tasks, preferably AI-supported.
• You measure what works: Individual cases are not proof, the same request can be answered differently twice. You check against a fixed collection of test requests and evaluate the change across all of them. The path from observation to change is short, many things go live on the same day.
• You are a sparring partner for Data and Engineering: For topics that lie deeper in the pipeline, you work closely with the departments. What you cannot implement yourself, you translate into well-founded requirements for Engineering and Data – with examples, numbers, and clear prioritization.
• New shops, new You will accompany partner onboarding, check what truly supports answer quality, and clearly state if a product range should not go live yet.
• Category management is your key: Categories, attributes, and facets play a role in what the assistant knows about a product – and because semantic similarity alone is not enough when it comes to dimensions, compatibility, or hard limits, you will get your hands dirty here yourself.

• You can demonstrate relevant professional experience in e-commerce, product management, or a comparable role in which you have truly driven a product forward and not just accompanied it.
• You are characterized by a strong hands-on mentality – you want to build things yourself and see them go live, rather than delegating and tracking them.
• You have worked practically with language models beyond occasional prompting: You have systematically improved prompts, compared results, and know where these systems fail – ideally with solutions like Claude, ChatGPT, or Gemini.
• Your intuition for user needs helps you recognize what lies behind a request and when an answer is only superficially suitable.
• You convince with an independent, results-oriented way of working and are eager to actively contribute your ideas to the team.
• Technically, you are adaptable: You can read an API response, understand a data structure, and are not afraid of SQL – you don't have to program, but you do have to get your hands dirty.
• Experience with retrieval, ranking, or semantic search, as well as with building evals and test sets is desirable.
• Knowledge of product data, feeds, taxonomies, and attribute systems is an advantage, as is experience with automation platforms like n8n, Make, or Python, and with MCP and agent connections.
• You have a fluent command of German and good English skills.

Your workday is yours
Work when you are most productive. Whether you're an early bird or a night owl: With us, you can flexibly adjust your working hours – in close coordination with your team. Working from another European country occasionally? No problem. Your laptop is your workplace.

Our culture is unique
Here, everyone sits on the same couch – regardless of title or role. We live team spirit digitally and in person, e.g., at gaming nights, workouts, movie nights, or team events.

We meet four times a year for Company Week
Working together in the office, pushing projects, enjoying pizza evenings, and concluding with a party.

In the heart of Hamburg
Centrally located between the main train station, Mönckebergstraße, and the Binnenalster – perfect for lunch dates, breaks by the Alster, or an after-work stroll.

Please submit your application documents exclusively through our online application form directly to our recruiting team.

Find Jobs in Germany on Arbeitnow