openProd extracts product values from supplier PDFs, spreadsheets and pasted text, maps them to the customer's attributes and units, and flags what it is unsure about. You will make that extraction more accurate and measure every change on real supplier files. What you will do
Improve how product values are extracted from supplier PDFs, spreadsheets and pasted text. Build evaluation sets from real supplier files and track accuracy from release to release. Improve how values are mapped to the customer's attributes, units and option lists. Decide when openProd should ask a person instead of guessing, and keep every value traceable to its source. Ship your work into the TypeScript API together with the product engineers. What we are looking for
Experience building features on top of large language models that run in production. Strong TypeScript or Python, and readiness to work in a TypeScript codebase. Experience extracting structured data from documents with prompts and output schemas. A habit of measuring: test sets, metrics and error analysis before and after a change. English at least for reading documentation, ideally fluent. Nice to have
Experience with PDF parsing, OCR or table extraction. Knowledge of product data, PIM systems or e-commerce catalogues. Experience with the Model Context Protocol (MCP). What we offer
Hybrid work and flexible hours. A modern office in Gdańsk near the airport, with a restaurant in the building. Private medical care and group life insurance. Multisport card and meal subsidies. English lessons with a native speaker. Internal training, trade fairs and conferences. How the process looks
Short call: A call of up to 30 minutes to see whether we are a good fit for each other. Test task: A task of up to 3 hours. You get our feedback within a week. Technical interview: We talk about your task and your experience. If we both say yes, we send you an offer.
