Crowdee is a small, research-driven team in Berlin building hybrid AI plus human content verification. We are not running an active hiring round at the moment. If you think you should be working with us anyway, write to us — we will read it and reply.
hello@crowdee.ai · Berlin, Germany · EU time zones preferred
We are not a generic AI startup. We are a research-rooted company that ships, with customers in newsrooms, broadcasters, and policy teams across Europe.
Crowdee grew out of the BMBF-funded news-polygraph project with DW, rbb, Fraunhofer IDMT, DFKI, and TU Berlin. We still treat each pipeline like a small research problem before it ships.
We do not believe pure AI gets verification right at scale. Every product decision we take involves both model performance and the workflows of the people who actually use the verdicts.
Our HQ is in Berlin-Mitte. We are happy to work fully remotely with people in EU time zones; we meet in person regularly when it matters.
We hire slowly because we care about who joins. We would rather wait for the right person than fill a seat — which is exactly why this page is what it is.
Send your application to hello@crowdee.ai with the subject "Initiative application". We read every message; expect a reply within roughly two weeks, faster when we can.
If your background sits anywhere near one of these areas, do not wait for a job ad — introduce yourself.
Synthetic-media detection, deepfake audio/video, NLP for veracity estimation, multimodal alignment, evaluation methodology. Publication track record welcome.
TypeScript, Next.js, Hono, Bun, Postgres, BullMQ on the application side; Python and inference infrastructure on the pipeline side. We value taste, clarity, and people who finish things.
Designing tasks for real human raters, running quality gates and inter-rater agreement, recruiting and managing the crowd, and translating editorial questions into structured judgements.
Selling into newsrooms, trust-and-safety teams, and policy units. Comfortable with a complex technical product, longer sales cycles, and a research-heavy story.
Fact-checkers, OSINT analysts, and editorial leads who can shape what an auditable verdict looks like — and what a good dossier needs to contain for a newsroom or a regulator.
We regularly take on working students and thesis collaborations in ML, NLP, signal processing, and product engineering. If your topic sits anywhere near what we do, write us.
Keep it short. We care more about clarity and craft than about cover-letter choreography.
One or two paragraphs: what kind of role you would want, what you have built or studied, what excites you about verification.
CV plus whatever shows your craft best — a GitHub profile, a paper, a portfolio, a deck, a notebook, a release post. Quality beats completeness.
Remote vs. Berlin, full-time vs. part-time, earliest start, salary expectation if you have one. Saying things plainly saves everyone time.
Your application data is processed solely to evaluate your interest in working with us, in line with our Privacy Policy. We delete applications we do not pursue within six months, unless you ask us to keep them on file.
If you have read this far, you are probably the kind of person we want to hear from anyway. Write us a paragraph and we will write you back.
