About the role
Lead Document AI Engineer - Intelligent Document Processing | Luxembourg
📍 Luxembourg (on-site) | 📅 ASAP | ⏳ 3-month project | 💼 Freelance
At PIXIE Luxembourg, we’re looking for a hands-on Lead Document AI Engineer for an important project with an insurance client in Luxembourg.
The client wants to analyse real customer emails and attachments, identify the documents received and turn them into structured data that operations teams can verify. If the system extracts an invoice amount, it needs to show where that amount came from and flag any inconsistency. If a document or field is missing, the output needs to say so.
This is an intensive three-month project, with a first usable Python package expected by the end of week three. You’ll lead the document processing and extraction workstream, write the code yourself and work alongside a Senior Python Engineer who owns the reference pipeline. The goal is a package that works on real emails in the client’s environment, not just a demo on sample documents.
🎯 Your mission
- Build the email and attachment analysis engine. Identify the documents in an email, classify them by type and apply the right extraction method to each one. Decide where OCR, layout analysis, rules or model-based extraction make sense.
- Make every extracted value traceable. Return structured JSON with the supporting passage for each value, the results of consistency checks and explicit flags for missing or uncertain information. Propose a request category and show what supports it.
- Prove that the extraction works. Use real historical emails to test document classification, field-level accuracy and source evidence. Investigate failures such as poor scans, conflicting values or incomplete attachments.
- Lead the technical delivery. Agree the interface with the reference-pipeline engineer, define what can credibly ship in three weeks and integrate both components into one tested, documented Python package that can later be exposed through FastAPI.
🔎 Who are we looking for?
- You’ve personally built and delivered structured data extraction across multiple document types, such as PDFs, scanned documents or forms.
- You’re a strong Python engineer who can make architecture decisions, implement them and debug the result on real documents.
- You can show how you linked an extracted value back to its source, validated the output and handled ambiguous or missing data.
- You can lead a short delivery cycle, make trade-offs when the data is messier than expected and work on-site in Luxembourg.
Experience in insurance or another document-heavy industry is a plus. What matters most is the extraction system you’ve already built and how you proved it worked.
🤝 Interested? Let’s connect !
👉 Apply directly through this LinkedIn job posting with your updated CV, earliest availability and expected daily rate. In a few lines, tell me which document types you handled, what you built yourself and how you measured extraction quality.
Any questions ? Feel free to message me directly.
#Freelance #Luxembourg #DocumentAI #IntelligentDocumentProcessing #Python #PIXIE